> Matt v. 1.2.1
=====

A Unix toolkit for analyzing genomic sequences
with focus on down-stream analysis of alternative splicing events

  Versions
       1.2.1
       1.2.0
       1.1.0
       1.0.0

       All & change log

Commands for:

Data analysis
• get_efeatures
• get_ifeatures
• cmpr_exons
• cmpr_introns
• cmpr_motif
• get_kmers_from_pwm
• get_pwm_hits
• get_pwm_prof
• get_regexp_hits
• get_regexp_prof
• perm_test
• plot_motif
• rna_maps
• rna_maps_cisbp
• test_cisbp_enrich
• test_pwm_enrich
• test_regexp_enrich

Data preparation
• add_cols
• add_rows
• add_val
• add_range
• chg_labs
• chk_tab
• col_calc
• col_uniq
• def_cats
• retr_geneids
• extr_scafids
• extr_seqs
• get_colnms
• get_cols
• get_match
• get_gcc
• get_ovl
• get_pwm
• get_rows
• get_sj
• conc
• get_seql
• get_seqs
• get_sss
• get_vast
• grp_calc
• mk_uniq
• perm_seqs
• prnt_tab
• rand_rows
• retr_rnaseq
• rm_cols
• rn_cols
• row_calc
• sort_tab
• stratify

Introduction
Download and installation
Tables: the central data structure
Documentation of all commands
Examples
Extending Matt: adding new commands
FAQ / Trouble shooting
How to cite

Introduction

Matt is a Unix command-line toolkit for analyzing genomic sequences with focus on the down-stream analysis of alternative splicing events. Being a POSIX-style command-line toolkit, Matt runs on the terminal of any Unix-like system.

Matt is modular: It comprises many commands, each of which tailored to one specific task. Combining commands allows the user to solve complex tasks: The whole is more than the sum of its parts.

The central data structure Matt works on is tables, where rows correspond to objects, e.g., exons, genes, etc., and columns correspond to features of these objects.

For performing a high-level analysis, the user of Matt usually goes through two phases.

Data preparation: Fit the output from other programs, e.g., for the estimation of inclusion levels of alternative splicing events from RNA-seq data, to the input format of Matt analyses. This might include:
- sub-selection of data
- defining groups of data to be compared
- changing group labels in tables
- adding information to tables, e.g., group labels, extract and add biological sequences
- stratifying data
- down-sampling data
Data analysis: run Matt analyses, e.g.:
- determining 75 exon features or 50 intron features
- comparing exons sets or intron sets for detecting discriminating features and generating a PDF report with p-values and box plots showing feature distributions
- testing motif enrichment in sequences with 297 CISBP-RNA binding motifs of RNA binding proteins
- generating motif RNA maps for any PWM/Perl REGEXP motif as high-quality PDF
- generating motif RNA maps automatically for all approx. 340 CISBP-RNA RBPs with IUPAC binding motif or for all ACGT k-mers of length 2 to 5
- visualizing positional densities of motif hits in sequences
- visualizing binding motifs of RNA binding proteins

Though the main focus of working with Matt are the high-level analyses, Matt offers commands for data preparation/table manipulation within the same framework, too. The motivation for this is to prevent users from implementing many highly similar scripts by themselves for standard preprocessing steps like defining groups of exons/introns to be compared with each other, extracting genomic sequences or stratifying data sets. In addition, users can use and combine with Matt any other tool for manipulating tables, e.g., other scripting languages like Awk, Sed, etc. or even LibreOffice Calc or Excel. Some analysis commands produce reports as PDF documents, PDF graphics, or web pages summarizing the results.

To make the work with Matt more efficient, many Matt commands are designed to be combined by piping. Exploiting this capability improves the work experience with Matt.

Matt provides comprehensive help messages on the terminal. To get an overview of all available commands, call

> matt

For getting the help message for COMMAND, call

> matt COMMAND

Each help message describes all arguments, e.g.,

matt COMMAND <ARGUMENT_1> [<ARGUMENT_2>] <ARGUMENT_3> ...

where <ARGUMENT> is a necessary argument, but [<ARGUMENT>] optional and can be omitted. The order of arguments always matters; changing the order might cause errors or false results.

This web page describes all commands and gives example calls. Both, this online documentation and the terminal help messages complement each other.

Download and installation

Create folder for installation
> mkdir matt
> cd matt
Download archive
> wget matt.crg.eu/matt.tar.gz
or
> curl -OL matt.crg.eu/matt.tar.gz
Decompress
> tar -xvzf matt.tar.gz
Run INSTALL script
> ./INSTALL
or
> source ./INSTALL
Add link to any PATH folder
> cd /path/to/PATH/folder
> ln -s /path/to/folder/matt/matt
Call Matt to get an overview of commands
> matt

Tables

Tab-separated plain text tables are the main data structure Matt works on. The motivations for making tables the basic data-structure are:

Compatibility: with the output of many programs for the estimation of inclusion levels of alternative splicing events that often output their results in form of tables, e.g., ASpli, IPSA MISO, rMATS, SANJUAN, SUPPA, VAST-TOOLS,
User-friendliness: most users are familiar with tables like spread sheets (LibreOffice Calc, Excel) and therefore can inspect tables easily with any spread sheet program
Clarity: combining all important pieces of information in a table with precise column names documents well the data
Flexibility: the order of columns in tables is never important; Having specified the names of columns to be used for a specific analysis, Matt will search for these columns in the table automatically. This allows the user to apply Matt to very different tables without the need of re-ordering their columns.
Exploiting piping: many Matt commands can be combined efficiently using piping as their main input and output are tables
Synergy: other powerful command-line tools like Sed or Awk can be applied to tables and, therefore, be used in combination with Matt

Tables must have headers with column names and column names must be unique, e.g., a table describing micro exons could look like this table t1.tab:

         EXON_ID    START   END     SCAFF  STRAND  SEQ      ENSEMBL_GID      GROUP
         ex1        123127  123132  chr1   +       GCTAGT   ENSG00000223972  up
         ex2        903861  903868  chr3   +       GCTCGATT ENSG00038737632  up
         ex3        4291861 4291863 chr5   -       GTC      ENSG00009875223  down
         ex4        9128712 9128717 chr12  -       CGCCCT   ENSG00002652352  down
         ex5        714614  714620  chrM   +       CGCTCGC  ENSG00083273821  up

It should be emphasized that Matt does not assume any specific column structure, rather does the user choose specific columns with data necessary for the command the user wants to apply. As a consequence, when calling Matt commands the user needs to provide as arguments the names of these columns. For example, a user might want to do a differential motif enrichment analysis with the micro-exons table above and would list columns SEQ and GROUP, or, when the user wants to retrieve splice-site sequences of these micro exons, the user would use the same table but list columns START, END, SCAFF, STRAND. The order of columns in tables is never important, as the user always need to refer Matt to specific columns by their names.

Not assuming specific columns in a specific order makes Matt flexible and applicable to almost any tab-separated table. For adding column names to tables yet without column names, e.g., like BED files, Matt offers the command add_header.

Users are free to choose any column names they prefer. Though, it is recommended to use capital letters for all COLUMN NAMES. Further, column names should contain only alpha-numeric characters, A-Z, 0-9, and '_'.

Generally, tables might contain any type of data representable as text, including long sequences. For improving the work experience with Matt, it is recommended that each column, whenever possible, should contain only one piece of information / value. Combining different values into one cell by concatenations, e.g., chr1:876123-877155, makes the single values, e.g., chromosome, start, end, inaccessible for Matt.

Matt can process tables with Unix new-lines and Windows new-lines. Tables with old-style Mac new-lines or Dos new-lines need to be adapted first; See command chk_nls. When working with Excel, please save tables in format Windows Text.

Documentation of Matt commands

This online documentation of Matt commands introduces all commands and provides examples and explanations. For more details on commands you might want to checkout the terminal help-messages which Matt offers for all commands.

For displaying an overview of all available commands, call

> matt

For getting a help message for COMMAND, call

> matt COMMAND

Each help message defines all arguments, e.g.,

matt COMMAND <ARGUMENT_1> [<ARGUMENT_2>] <ARGUMENT_3> ...

where <ARGUMENT> is a necessary argument, but [<ARGUMENT>] optional and can be omitted.

Generally, one might categorize all Matt commands into commands for

Data preparation/Table manipulation for analysis (top of left panel)
Data analysis (top of left panel)

If you are primarily interested in performing analyses, you might want to make yourself familiar with the commands for analyses first, and later learn about the commands for data preparation.

Table categorizing all Matt commands into eight groups

1. Import data / check table

chk_nls visualize and check newlines in table

chk_tab check if table is Matt conform, i.e., all rows have same number of columns

get_pwm read-in position weight matrix model from text file

get_sj read-in output table with skipped exons from SANJUAN program

get_vast read-in output table with alternative splicing events from VAST-TOOLS program

2. Retrieve data from table

col_uniq list all distinct values of a column

get_colnms get column names of table

get_cols extract entire columns from table

get_match match entries across two tables via IDs

get_ovl get overlap, i.e., events (rows) which are present in several tables

conc construct artificial row ids by concatenating specific values across rows

get_rows extract entire rows from table, which fulfill constraints

rand_rows randomly down-sample rows from table

3. Manipulate table

add_cols add columns to table

add_header add header to table yet without header

add_range add column with range of values increasing for each row

add_rows add rows to table

add_val add one new column with constant value to table

chg_labs in a column with categorical values (group labels), change these labels

def_cats define group labels for items (rows) wrt. their features

rm_cols remove columns from a table

rn_cols rename column names in header of table

4. Retrieve sequences

extr_seqs output in FASTA format sequences from a column of a table

get_seqs retrieve sub-sequences (e.g., exons, genes) from a FASTA file

5. Analyze sequences

cmpr_motif generate a comparative motif plot (PDF) for comparing two motifs of the same length

get_efeatures retrieve 75 features for exons

get_gcc determine the GC content for sequences in a column of a table

get_ifeatures retrieve 50 features for introns

get_pwm_hits search in sequences for hits with a motif position weight matrix model

get_pwm_prof visualize positional distributions of PWM hits across sequences

get_regexp_hits search in sequences for hits with Perl regular expressions

get_regexp_prof visualize positional distributions of REGEXP hits across sequences

get_seql get length of sequences given in a column of a table

get_sss get splice-site scores (MaxEnt models) for splice sites

plot_motif generate a motif plot as PDF (similar to sequence logos)

test_pwm_enrich test motif enrichment in two sets of sequences with motif position weight matrix

test_regexp_enrich test motif enrichment in two sets of sequences with motif regular expression

6. Maths and statistics

row_calc apply mathematical computations along rows

col_calc apply mathematical computations along columns

perm_test permutation test for numeric, categorical, co-occurrence features

7. General purpose

retr_geneids retrieve gene IDs from GTF for, e.g., genes, exon, introns

extr_scafids get overview of chromosome/scaffold IDs from FASTA, FASTQ, GTF files

get_kmers_from_pwm get top k-mers most likely under a given position weight matrix model

mk_uniq make unique a table, i.e., neglect rows with replicated values

perm_seqs randomly permute nucleotides within sequences

prnt_tab print a table nicely to screen

retr_rnaseq retrieve RNA-seq data seqs from GEO with GEO accession numbers

sort_tab sort a table according to values along column

stratify stratification of two data sets wrt. a numerical feature

8. High-level analyses

cmpr_exons comparing exon groups: extract 75 exon features and determine discriminating features

cmpr_introns comparing intron groups: extract 50 intron features and determine discriminating features

rna_maps for groups of exons / introns: generate PDF plots with motif RNA maps

rna_maps_cisbp motif RNA maps for all CISBP-RNA motifs, or 2 to 5-mers

test_cisbp_enrich for groups of sequences: check for motif enrichment with all 297 CISBP-RNA binding motifs

> matt chk_nls

Description: Use this command to print a table to screen making visible new-lines. This helps to investigate which new-line characters are present in the table and how the table would need to be adapted to be processable by Matt. This might be necessary for tables with old-style Mac or Dos new-lines.

Example: With t1.tab being the example table from above

> matt chk_nls t1.tab

prints this table making visible new-lines.

> matt chk_tab

Description: Use this command to check if a given table is processable by Matt. To be so, all rows of a table must have the same number of fields. This command lists row-number of rows with a wrong number of fields or reports that the table is correctly formatted.

Example: With t1.tab being the example table from above

> matt chk_tab t1.tab

reports that this table is correctly formatted and, hence, processable by Matt.

> matt get_pwm

Description: Use this command to

read-in a position weight matrix (PWM) given by a table
learn a PWM from aligned sequences of identical length.

and turn it into the standard Matt format for PWMs. The first case is helpful when you have down-loaded a PWM or when you have extracted it from some document and want to use it with Matt. Matt tries to understand the structure of the table, e.g., if rows or columns correspond to positions. The second case is helpful when you have short sequences of identical length, e.g., splice sites or branch points, from which you want to learn a PWM (get the relative frequencies of letters at each position) and use it with Matt.

Example: With M037_0.6.txt being the human MBNL1 PWM from CISBP RNA

> matt get_pwm M037_0.6.txt 0.001 A,C,G,T

prints this MBNL1 PWM with minimal probability 0.001 instead of 0 and alphabet A,C,G,T instead of A,C,G,U. For re-defining the alphabet, you could also directly edit the text file M037_0.6.txt. To save this PWM in Matt format for later use with Matt, re-direct the output into a file.

> matt get_pwm M037_0.6.txt 0.001 A,C,G,T > hsa_mbnl1_pwm.tab

> matt get_sj

Description: This command is tailored to the result table of SANJUAN (a program for estimating inclusion levels of alternative splicing events from RNA-seq data). These tables describe exons and already contain their start and end-coordinate, chromosome, and strand. With this command, you'll add Ensembl gene ids to the table, which are necessary for subsequent Matt analyses.

Example: SANJUAN outputs differentially spliced exons in a table called CEs_clean.txt.

> matt get_sj CEs_clean.txt hg19 > exons_from_sj.tab

will output a table like CEs_clean.txt with an additional column GENEID_ENSEMBL. It will duplicate rows (exons) for which SANJUAN did output several GENE_SYMBOLs; one row for each of the reported genes with its gene name and Ensembl gene id.

> matt get_vast

Description: This command is tailored to the result tables of VAST-TOOLS. Use this command to transform the output table from VAST-TOOLS into Matt format and extract sub-sets of reported alternative splicing events. VAST-TOOLs reports skipped exons, retained introns, alt3, alt5 events in one output table together with estimated PSI values across several data sets. Matt allows you to retrieve form this table events of specific types (skipped exons, retained introns, Alt3, Alt5) or events that have a PSI value within a user specified range across all samples. Before you run a high-level analysis you might need to defined groups of events (e.g. up-regulated vs down-regulated skipped exons) with Matt function def_cats applied to the result table of get_vast. More information on the format of the VAST-TOOLs output table can be found here.

Example: With vts_out.tab being a final results table from VAST-TOOLS combine command

> matt get_vast vts_out.tab -minqab LOW -minqglob N -complex IR,IR-S,IR-C 
     -a kd1,kd2 -b ctr1,ctr2 > ir_events.tab

extracts all intron retention events from vts_out.tab together with detailed information on PSI values (min, max, mean) across all samples, across samples -a kd1,kd2, across samples -b ctr1,ctr2, and dPSIs for the comparison of samples -a vs samples -b. In order to be considered in these calculations, PSI values need to have a minimum quality flag of LOW in samples -a and -b and N across all other samples. If for certain events no PSI values are left, NA is output.

The user can augment the result table (e.g. ir_events.tab) with gene IDs extracted from any given gene annotation (e.g. GTF file Hsa19.gtf with gene ids in field gene_id) by

> matt get_vast vts_out.tab -minqab LOW -minqglob N -complex IR,IR-S,IR-C 
     -a kd1,kd2 -b ctr1,ctr2 -gtf Hsa19.gtf -f gene_id > ir_events.tab

The output table ir_events.tab will have one additional column GENEID with the extracted gene IDs. Of course do the chromosome annotations in the VAST-TOOLs output need to match to those in the GTF file.

Generally, the workflow is as follows: pre-process the output table of VAST-TOOLS for extracting PSI values with get_vast, define groups of events to be analyzed with def_cats, and eventually run a high-level analysis.

> matt col_uniq

Description: Use this command to get all distinct values of a column together with their number of occurrences. This command should be applied only to columns with categorical data.

Example: The ir_events.tab containing intron-retention events from some VAST-TOOLs output obtained with command get_vast. This table has a column COMPLEX which contains for each event the sub-types (IR, IR-S, IR-C) as defined by VAST-TOOLS:

> matt col_uniq ir_events.tab COMPLEX

prints the three sub-types and their numbers of occurrences in table ir_events.tab.

> matt get_colnms

Description: Use this command to get all column names and their indices from a table. This command is useful to investigate the structure of tables and the kind of data in their columns.

Example: With t1.tab being the table from above

> matt get_colnms t1.tab

prints the column headers EXON_ID, START, END, SCAFF, STRAND, SEQ, ENSEMBL_GID, GROUP and their indices.

> matt get_cols

Description: Use this command to retrieve one or several columns from a table.

Example: With t1.tab being the table from above

> matt get_cols t1.tab START END

prints the entire columns START and END from table t1.tab.

You might want to redirect the output into a file to construct a new table consisting of the extracted columns

> matt get_cols t1.tab START END > new_table.tab

> matt get_match

Description: For a given table with entities, find matching values from a second table. E.g., assume you have a table 1 with a column with gene ids and another table 2 with gene ids and gene names, then you can extract gene names from table 2 which match to the gene ids in table 1.

Example: With t1.tab being the table from above and t2.tab describing a mapping GENE_ID→GENE_NAME with columns

GENE_ID
GENE_NAME

> matt get_match t1.tab GENE_ID t2.tab GENE_ID GENE_NAME

prints a table with one column GENE_NAME. Each row of this table corresponds to a row of t1.tab and it contains the GENE_NAME for the corresponding GENE_ID of t1.tab. Piping allows the user to concatenate both with command add_cols and augment table t1.tab.

> matt get_match t1.tab GENE_ID t2.tab GENE_ID GENE_NAME | matt add_cols t1.tab -

> matt get_ovl

Description: Use this command to retrieve IDs of rows which occur in several tables. This command might be helpful if you want to determine the overlap of different data sets.

Example: Table ir_events.tab contains intron-retention events and has a column EVENT with event IDs. Table ir_events_reference.tab contains reference intron-retention events with column EVENT containing their event IDs. Then

> matt get_ovl ir_events.tab EVENT ir_events_reference.tab EVENT > common_ids.tab

writes a table with column OVERLAP with event ids that occur in table ir_events.tab and ir_events_reference.tab into file common_ids.tab. To extract all rows from table ir_events.tab for these events, you can apply get_rows as follows

> matt get_rows ir_events.tab EVENT]common_ids.tab,OVERLAP[ > common_ir_events.tab

> matt conc

Description: Use this command for concatenating entries of several columns. This command might be useful for constructing artificial row IDs, i.e., event IDs which might be necessary, e.g., for get_ovl if the table not yet contains event IDs.

Example: With t1.tab being the table from above

> matt conc t1.tab EVENT_ID '_' START END SCAFF STRAND

prints a table with one column EVENT_ID which contains in each row an event ID by concatenating START_END_SCAFF_STRAND. With command add_cols you might add this column with event IDs to the table t1.tab.

> matt conc t1.tab EVENT_ID '_' START END SCAFF STRAND | matt add_cols t1.tab -

Other separation symbols like the empty string '' or white space ' ' can be specified.

> matt get_rows

Description: Use this command to retrieve rows (items) from a table. Filters allow the user to select specific rows, i.e., rows which have specific values (numeric or categorical) in some of the columns, or whose vales are empty (empty string).

Example: With t1.tab being the table from above

> matt get_rows t1.tab START[1,5000000] '!STRAND]+['

prints the rows from t1.tab with START in 1 to 5 Mio and STRAND is NOT +.

Constraints can be:

LENGTH[0,5] : numeric values in column LENGTH must be in 0 to 5 (inclusive)
LENGTH[0,5,10,15] : numeric values in column LENGTH must be in 0 to 5 or in 10 to 15
GROUP]g1,g2[ : categorical values in column GROUP must be either g1 or g2
GROUP]table.tab,column_name[ : categorical values in column GROUP must be in column column_name of table table.tab (useful if you want to specify many admissible categorical values stored in a table)
GROUP][ : categorical value in column GROUP must be empty string
COLUMNA==COLUMNB: entries in both columns must be equal
'!LENGTH[0,5]' : negation; numerical value in column LENGTH not in 0 to 5
'!COLUMNA==COLUMNB' : negation; entries in both columns must not be equal
constraints can be combined

> matt rand_rows

Description: Use this command to randomly draw rows from a table. This command might be useful when you want to sub-sample data sets.

Example: The table ir_events.tab contains intron-retention events:

> matt rand_rows ir_events.tab 500 3628929065

prints 500 randomly drawn (seed 3628929065) rows (without replacement) from ir_events.tab. You might want to save these events into a file for later use.

> matt rand_rows ir_events.tab 500 3628929065 > 500_ir_events.tab

> matt add_cols

Description: Use this command to add columns to existing tables. Columns to be added might be newly constructed or retrieved from another table with get_cols.

Example: With t1.tab being the table from above and having constructed row IDs with conc like

> matt conc t1.tab EVENT_ID _ START END SCAFF STRAND > event_ids.tab

doing

> matt add_cols t1.tab event_ids.tab

will add the column EVENT_ID from event_ids.tab to the right end of table t1.tab.
Combining both commands using piping is more efficient.

> matt conc t1.tab EVENT_ID _ START END SCAFF STRAND | matt add_cols t1.tab -

Important: Column names must be unique, i.e., you cannot join a column to a table if this table has already a column with the same column name. To rename columns, you might use rn_cols. It can be efficient to use piping here, e.g.:

> matt rn_cols a.tab COLA:RENAMED_COLA | matt add_cols b.tab -

> matt add_header

Description: Use this command to add column names (entire table header) to a table yet without header to make it processable with Matt.

Example: exons.bed being a BED file with the first three mandatory fields only

> matt add_header exons.bed chrom chromStart chromEnd

adds three column names as header to the file exons.bed directly without printing it to screen.

> matt add_range

Description: Use this command to add a column with a range index increasing for each row.

Example: With t1.tab from above

> matt add_range t1.tab ID

prints t1.tab with extra column ID with index starting at 0 and increasing in each row.

> matt add_rows

Description: Use this command to add rows to existing tables. You might want to add rows of one table to another table, e.g., to join data sets into a single table.

Example: With t1.tab being the table from above

> matt add_rows t1.tab t1.tab

will add all rows from t1.tab to the bottom of t1.tab, which does not look very meaningful.
Important: Both tables must have the same number of columns with the same column names, but their order can vary as Matt will re-order columns automatically if necessary.

> matt add_val

Description: Use this command to add one column with a constant value across all rows to a existing table. You might use this command to add a column containing some kind of group IDs (e.g., up_regulated) to tables which you later join into one table.

Example: With t1.tab being the table from above

> matt add_val t1.tab DATE 22.02.2017

will print table t1.tab to screen with a additional column DATE containing the value 22.02.2017 in all rows.

> matt chg_labs

Description: Use this command to change labels (categorical entries) in a column of a table. You might use this command if you have a column with group IDs and want to change these IDs / labels.

Example: With t1.tab being the table from above

> matt chg_labs t1.tab EXON_ID ex1:ex11 ex2:ex22

will print table t1.tab to screen where in column EXON_ID ex1 is changed for ex11, and ex2 for ex22.

> matt def_cats

Description: Use this command (define categories) for creating a new column with some kind of group IDs. Generally, rows of table corresponds to events, while columns correspond to their features. Hence, you define group IDs (one for each row) wrt. entries in the feature columns using the same constraint syntax as used with command get_rows.

This command is important as many analyses essentially compare some groups of items (e.g. groups of exons, introns, sequences, or more), hence, the user needs to defined the groups to be compared.

Example: With t1.tab being the table from above

> matt def_cats t1.tab GROUP2 'g1=START[0,5000000] STRAND]+[' 
     'g2=START[0,5000000] !STRAND]+[' 'g3=START[5000001,10000000]'

will output a table with column GROUP2 with group IDs g1, g1, g2, g3, g1 categorizing each micro exon in table t1.tab wrt. the defined constraints.

> matt rm_cols

Description: Use this command to remove columns from a table changing the original table.

Example: With t1.tab being the table from above where we add a column with

> matt add_val t1.tab DATE 22.02.2017 > t2.tab

doing

> matt rm_cols t2.tab DATE STRAND

will print a table with columns DATE and STRAND to screen and remove these columns from the original t2.tab.
Be careful: The original table will be changed by removing columns. This might lead to undesired data loss.

> matt rn_cols

Description: Use this command to rename columns of tables. You might use this command before you want to join by-column two tables which have some column names in common. Tables must not contain column names several times. Each column needs to have a unique name.

Example: With t1.tab being the table from above

> matt rn_cols t1.tab SCAFF:SCAFFOLD STRAND:STRAND2

will print t1.tab to screen with column names SCAFFOLD instead of SCAFF and STRAND2 instead of STRAND.

> matt extr_seqs

Description: Use this command to output in FASTA format sequences from a table. You might use this command if you want to run other software on sequences which you have obtained with Matt. Often, software for analyzing sequences expect FASTA input format.

Example: With t1.tab being the table from above

> matt extr_seqs t1.tab SEQ EXON_ID

will print the sequences from column SEQ into FASTA format where values in column EXON_ID will be used as sequence IDs.

If the table not yet contains a column with kind of sequence IDs, you might want to use conc to construct them.

> matt get_seqs

Description: Use this command to retrieve genomic sub-sequences from FASTA files. Usually, you'll have a table specifying genomic coordinates (e.g. START, END, SCAFFOLD/CHROMOSOME, STRAND) and you want to retrieve genomic sequences with these coordinates (e.g., sequences of exons, introns, genes, splice sites or any other genomic entity) from a FASTA file (which contains all sequences of the entire genome). The command get_seqs has three modes making it easy to retrieve different genomic sequences wrt. to the coordinates.

Schematic view of positions of START and END coordinates and the four length arguments for mode 3.

The three modes make it easy to retrieve:

sequences from START to END, including START and END position
two sequences per START-END coordinate pair (START and END exclusive)
1. up-stream sequence of length L1 of region START-END, i.e., from START-L1 to START-1 (on + strand)
2. down-stream sequence of length L2 of region START-END, i.e., from END+1 to END+L2 (on + strand)
two sequences per START-END coordinate pair (START and END inclusive)

up-stream sequence around START, i.e., from START-L1 to START+L2-1 (on + strand)
down-stream sequence around END, i.e., from END-L3+1 to END+L4 (on + strand)

Sequences from - strand get reverse-complemented automatically and up/down-stream sequences come from the up/down-stream regions wrt. START end END coordinates.

Example: With t1.tab being the table from above

> matt get_seqs t1.tab START END 20 10 SCAFF STRAND Hsa_genome.fa

will extract for all micro exons in t1.tab the up-stream intronic sequences of length 20 and the down-stream intronic sequences of length 10 and print these sequences as a table with columns SEQ_UP and SEQ_DOWN. The scaffold/chromosome IDs in column SCAFF of table t1.tab must be the same as used in Hsa_genome.fa.

Using piping, you can add the retrieved sequences directly to the table t1.tab as get_seqs outputs the sequences as a two-column table.

> matt get_seqs t1.tab START END 20 10 SCAFF STRAND ...
       ... Hsa_genome.fa | matt add_cols t1.tab -

> matt cmpr_motif

Description: Use this command to plot a PDF for visually comparing two motifs (position weight matrices) of the same length. Motifs can be given either by tables describing position weight matrices (get_pwm) or by tables containing sequences (aligned, all of same length) from which Matt will estimate position weight matrices.

Example: With hsa_mbnl1_pwm.tab and hsa_a2bp1_pwm.tab being two PWMs retrieved from CISBP-RNA, both being of length 7,

> matt cmpr_motif hsa_mbnl1_pwm.tab hsa_a2bp1_pwm.tab mbnl1_vs_a2bp1.pdf

generates the PDF mbnl1_vs_a2bp1.pdf which shows the motif plots (plot_motif) for both PWMs and in the center a diagram showing at each position the log2-ratio of probabilities of each nucleotide.

> matt get_efeatures

Description: Use this command for retrieving 75 features of interest for exons. Exons need to be described by a table with basic information like their genomic coordinates and a gene ID of genes the exons belong to. If the table does not yet contain gene IDs, you might use the command retr_geneids for extracting gene IDs from any GTF file for given genomic events, like exons, introns, genes.

Overview of features

Motivation for including features

GENE BIOTYPE: Splicing of exons and introns in coding vs non-coding genes might underlie different mechanisms.
LENGTH: Exon and intron length, as well as the relation of exon length and the lengths of its neighbor introns, and intron length to the lengths of its neighbor exons might indicate different splicing mechanisms (e.g., exon vs intron definition, Amit et al., DOI: 10.1016/j.celrep.2012.03.013), or categories of exons like micro-exons or long introns.
GC CONTENT of exons, introns, neighboring exons and introns and their ratios, and GC contents of splice sites: The GC content together with the length and lengths ratios might indicate different splicing mechanisms (e.g., exon vs intron definition, Amit et al., DOI: 10.1016/j.celrep.2012.03.013)
The following BP and splice-site features: They are proxies for the binding specificity respect to the U2 snRNPs and other core spliceosomal components. The lower this binding specificity is, the less the intron/exon will be recognized.
SF1 BINDING SCORES: Is an approximation of the branch-point strength thought of impacting the efficiency of splicing.
MAXENT SCORES: of splice sites are an approximation of the efficiency of splice sites (Yeo et al., DOI: 10.1089/1066527041410418)
DISTANCE PREDICTED BRANCH POINTS TO 3SS, NUMBER OF PREDICTED BRANCH POINTS, SEQUENCE SCORE OF BRANCH POINT SEQUENCE, PYRIMIDINE CONTENT, LENGTH OF PYRIMIDINE TRACT, OFFSET OF PYRIMIDINE TRACT FROM 3SS, PYRIMIDINE TRACT SCORE: are used for the classification of branch points by Corvelo et al., DOI: 10.1371/journal.pcbi.1001016; Thus they are potentially interesting features to be considered when studying and comparing introns or neighbor exons.
MEDIAN EXON/INTRON NUMBER: of transcripts a specific exon/intron occurs in; Might indicate if exons/introns preferentially occur in transcripts with few or many exons/introns.
(RELATIVE) RANK: of exon or intron. Might reveal if exons/introns preferentially occur at the start, interior, or end of transcripts.
PROPORTION OF TRANSCRIPTS: where exon/intron occurs as first, internal, or last exon/intron, or within UTRs. Again, might indicate bias of the location of exons/introns in transcripts.
CO-OCCURRENCES: of exons/introns in same transcripts. Might indicate of studied exon/introns tend to co-occur in transcripts.

cmpr_exons

Examples:

> matt get_efeatures explain

t1.tab

> matt get_efeatures t1.tab START END SCAFF STRAND ENSEMBL_GID ...
      ... Hsa19.gtf Hsa19.fa Hsap  > efeatures.tab

If only one or a few features are of interest, users can apply get_cols and extract specific feature columns only.

> matt get_efeatures t1.tab START END SCAFF STRAND ENSEMBL_GID ...
      ... Hsa19.gtf Hsa19.fa Hsap  | matt get_cols - EXON_LENGTH EXON_GCC > efeatures.tab

It is recommended to use GTF files with genome annotation from Ensembl, as Matt is tailored to their structure.

> matt get_gcc

Description:

Example:

t1.tab

> matt get_gcc t1.tab SEQ

Any letter not A, C, G, T, U, a, c, g, t, u will be completely neglected from the computation of the GC content.

> matt get_ifeatures

Description:

retr_geneids

Overview of features

get_ifeatures

cmpr_introns

Examples:

> matt get_ifeatures explain

> matt get_ifeatures introns.tab START END SCAFF STRAND ENSEMBL_GID ...
      ... Hsa19.gtf Hsa19.fa Hsap > ifeatures.tab

If only one or a few features are of interest, users can apply get_cols and extract specific feature columns only.

> matt get_ifeatures t1.tab START END SCAFF STRAND ENSEMBL_GID ...
      ... Hsa19.gtf Hsa19.fa Hsap  | matt get_cols - INTRON_LENGTH INTRON_GCC > ifeatures.tab

> matt get_pwm_hits

Description:

Example:

hsa_mbnl1_pwm.tab

> matt get_pwm_hits seqs.tab SEQ ID hsa_mbml1_pwm.tab 10 single INFER_BGPWM

> matt get_pwm_prof

Description:

Example:

SEQ_UP: up-stream intronic sequences (150 nt)
SEQ_DOWN: down-stream intronic sequences (150 nt)
DATASET: all exons are classified into non-differentially spliced (ndiff) and differentially spliced (down) and this column contains the group IDs: ndiff and down

            NAME    FILE        BG            THRESH
            A2BP1   a2bp1.tab   INFER_BGPWM   5
            MBNL1   mbnl1.tab   INFER_BGPWM   5

get_pwm

> matt get_pwm_prof exons.tab SEQ_UP RL SEQ_DOWN LR ...

         ... pwms.tab NAME FILE THRESH BG DATASET[down,ndiff]

estimate a background PWM by taking the frequencies of the four nucleotides A,C,G,T across all sequences in column and group
predict hits at all sequence positions with P_fgPWM(seq)/P_bgPWM(seq)≥5
estimate a positional density of predicted hits with a Normal kernel considering all hits
estimate a positional cumulative density considering all hits
estimate a positional density with Normal kernel considering only the first hit per sequence
estimate a positional cumulative density considering only the first hit per sequence

> matt get_pwm_prof exons.tab SEQ_UP RL SEQ_DOWN LR ...

         ... pwms.tab NAME FILE THRESH BG DATASET[down,ndiff]

Example website with results:

Matt: Positional distributions of hits of motif PWMs

Data file: exons.tab
Categories (number of sequences): down ( 868 ), ndiff ( 661 )
Sequence columns selected (plotting parameter): SEQ_UP ( RL ), SEQ_DOWN ( LR )
File with information on motif PWMs and background models: pwms.tab
Date: Feb 27, 2017

Ordered Results

Ordering from top to bottom in each column:

Reference group:

none

Densities all positions	Cumulative densities all positions	Densities first positions	Cumulative densities first positions
MBNL1	MBNL1	MBNL1	MBNL1
A2BP1	A2BP1	A2BP1	A2BP1

Plots

A2BP1 - a2bp1.tab

Back to Ordered Results

MBNL1 - mbnl1.tab

Back to Ordered Results

> matt get_regexp_hits

Description:

Example:

> matt get_regexp_hits seqs.tab SEQ ID 'ACCG|G[GC]C|GCCT' single

> matt get_regexp_prof

Description:

Example:

SEQ_UP: up-stream intronic sequences (150 nt)
SEQ_DOWN: down-stream intronic sequences (150 nt)
DATASET: all exons are classified into non-differentially spliced (ndiff) and differentially spliced (down) and this column contains the group IDs: ndiff and down

                      NAME    REGEXP
                      MBNL1   [GC][CT]TT[GA]C  
                      A2BP1   [TA]GCATG[CA]|TACATA

> matt get_regexp_prof exons.tab SEQ_UP RL SEQ_DOWN LR ...

         ... regexps.tab DATASET[down,ndiff]

predict hits (exact matches) for regular expression
estimate a positional density of hits with Normal kernel considering all hits
estimate a positional cumulative density of hits considering all hits
estimate a positional density with Normal kernel considering only the first hit per sequence
estimate a positional cumulative density considering only the first hit per sequence

> matt get_regexp_prof exons.tab SEQ_UP RL SEQ_DOWN LR ...

         ... regexps.tab DATASET[down,ndiff]

Example website with results:

Matt: Positional distributions of hits of REGEXPs

Data file: exons.tab
Categories (number of sequences): down ( 868 ), ndiff ( 661 )
Sequence columns selected (plotting parameter): SEQ_UP ( RL ), SEQ_DOWN ( LR )
K or file with information on REGEXPS: regexps.tab
Date: Feb 27, 2017

Ordered Results

Ordering from top to bottom in each column:

Reference group:

none

Densities all positions	Cumulative densities all positions	Densities first positions	Cumulative densities first positions
A2BP1	A2BP1	A2BP1	A2BP1
MBNL1	MBNL1	MBNL1	MBNL1

Plots

MBNL1 - [GC][CT]TT[GA]C

Back to Ordered Results

A2BP1 - [TA]GCATG[CA]|TACATA

Back to Ordered Results

> matt get_seql

Description:

Example:

t1.tab

> matt get_seql t1.tab SEQ

> matt get_sss

Description:

maximum-entropy models

get_seqs

Example:

> matt get_sss introns.tab 3SS_SEQ 5SS_SEQ

Using piping allows the user to extract the sequences and estimate the splice site strength at once. E.g., assume table introns.tab has columns START, END, SCAFF, STRAND describing human introns:

> matt get_seqs introns.tab START END 3 6 20 3 SCAFF STRAND Hsa19.fa | matt ... 
   ... get_sss - SEQ_UP SEQ_DOWN

> matt plot_motif

Description:

get_pwm

Example:

human NOVA2 PWM

> matt plot_motif human_nova2_pwm.tab

> matt test_pwm_enrich

Description:

Example:

Sequence data:

SEQ: containing sequences
GROUP_ID: containing group ids (cntr, up, down)

PWM specification:

NAME: containing short names for the motifs
FILE: containing the path to file with table defining the PWM (in Matt format: see get_pwm)
THRESH: containing the threshold to be applied for predicting hits with this PWM, e.g., 10 (more details get_pwm_hits)

                NAME   FILE                THRESH
                pwm1   /path/to/pwm1.tab   10
                pwm2   /path/to/pwm2.tab   10
                pwm3   /path/to/pwm3.tab   10

> matt test_pwm_enrich exons.tab SEQ GROUP_ID[up,cntr,down,cntr] single ...
      ...  pwms.tab NAME FILE THRESH INFER_BGMODEL quant 10000 > results.tab

estimate background model bm1 (homogeneous Markov model of order 0) from sequences of the foreground data set, here group up
estimate background model bm2 from sequences of the background data set, here group cntr
predict hits with pwm1 in sequences of group up (ratio pmw1/bm1≥10)
predict hits with pwm1 in sequences of group cntr (ratio pmw1/bm2≥10)
apply a permutation test with 10000 iterations (repeats) and determine p-value

Test for positional differences of hits:

cntr

the entire sequences as given
only the first 1/3 part of the sequences
only the internal 1/3 part of the sequences
only the last 1/3 part of the sequences

Output:

SIGNIFICANT_RESULT_ALL: enrichment considering entire sequences
SIGNIFICANT_RESULT_FIRST: enrichment considering only the first 1/3 of the sequences
SIGNIFICANT_RESULT_INTERNAL: enrichment considering only the internal 1/3 of the sequences
SIGNIFICANT_RESULT_END: enrichment considering only the last 1/3 of the sequences

Number and length of sequences:

Avoid false positives:

cntr

Change test statistic of the enrichment analysis: number of hits vs. number of sequence with hits

the number of hits across sequences per group of sequences
the number of positive sequences containing hits per group

argument <MODE>

> matt test_regexp_enrich

Description:

test_pwm_enrich

get_regexp_hits

Example:

Sequence data:

SEQ: containing sequences
GROUP_ID: containing group ids (cntr, up, down)

Specification of Perl regular expressions / patterns:

NAME: containing short names for the motifs/patterns, e.g, the corresponding binding protein
REGEXP: containing the Perl regular expressions

                NAME   REGEXP
                protA  ACCGT|ACCGA|ACCAA
                protB  AC[CGT]..C{2,5}GT

> matt test_regexp_enrich exons.tab SEQ GROUP_ID[up,cntr,down,cntr] single ...
      ...  regexps.tab NAME REGEXP quant 10000 > results.tab

search hits (exact matches) with regular expression protA in sequences of group up
search hits (exact matches) with regular expression protA in sequences of group cntr
apply a permutation test with 10000 iterations (repeats) and determine p-value

Test for positional differences of hits:

cntr

the entire sequences as given
only the first 1/3 part of the sequences
only the internal 1/3 part of the sequences
only the last 1/3 part of the sequences

Output:

SIGNIFICANT_RESULT_ALL: enrichment considering entire sequences
SIGNIFICANT_RESULT_FIRST: enrichment considering only the first 1/3 of the sequences
SIGNIFICANT_RESULT_INTERNAL: enrichment considering only the internal 1/3 of the sequences
SIGNIFICANT_RESULT_END: enrichment considering only the last 1/3 of the sequences

Number and length of sequences:

Avoid false positives:

cntr

Change test statistic of the enrichment analysis: number of hits vs. number of sequence with hits

the number of hits across sequences per group of sequences
the number of positive sequences containing hits per group

argument <MODE>

> matt row_calc

Description:

Example:

t1.tab

> matt row_calc t1.tab 'LENGTH=END-START+1'

> matt row_calc t1.tab 'LENGTH=END-START+1' | matt add_cols t1.tab -

> matt row_calc t.tab 'MEAN=log((START+END+LENGTH)/3)'

> matt col_calc

Description:

Example:

         A    B    V1    V2
         a    1    1     2
         a    2    1     3
         b    2    1     4
         a    1    1     5

> matt col_calc t2.tab sum V1,V2

         V1_SUM    V2_SUM
         4         14

> matt col_calc t2.tab sum V1,V2 A

         A    V1_SUM    V2_SUM
         a    3         10
         b    1         4

> matt col_calc t2.tab sum V1,V2 A,B

         A    B     V1_SUM    V2_SUM
         a    1     2         7
         a    2     1         3
         b    2     1         4

> matt perm_test

Description:

numerical
categorical
co-occurrences

Example:

get_efeatures

LENGTH: length of exon
GCC: GC content of exon
GENE_TYPE: coding, non-coding
IS_IN_TRS: colon-separated list of transcript ID each exon occurs in
DATASET with group IDs down, ndiff

> matt perm_test exons.tab DATASET[down,ndiff] 100000 ...

           ... -num LENGTH,GCC -card GENE_TYPE -cooc IS_IN_TRS

Co-occurrences:

> matt retr_geneids

Description:

Example:

> matt retr_geneids exons.tab START END SCAFFOLD STRAND Hsa19.gtf -f gene_id

> matt extr_scafids

Description:

Example:

> matt extr_scafids Hsa19_gDNA.fasta FASTA

> matt get_kmers_from_pwm

Description:

Example:

mbnl1.tab

get_pwm

> matt get_kmers_from_pwm mbnl1.tab 5 5

          ID  KMER     PWM_LOGSCORE       CUMULATIVE_PERCENT_BINDING  PWM_NAME
          1   AGCTTGC  -1.98607349885094  0.137233215315936           mbnl1
          2   TGCTTGC  -2.09775518556902  0.259964844297944           mbnl1
          3   GGCTTGC  -2.38376620889964  0.352167512560293           mbnl1
          4   CGCTTGC  -2.49208691693447  0.434904633342322           mbnl1
          5   AGCTTGG  -3.67966981232818  0.460135937867991           mbnl1

> matt get_kmers_from_pwm mbnl1.tab 0.5 5

          ID  KMER     PWM_LOGSCORE       CUMULATIVE_PERCENT_BINDING  PWM_NAME
          1   AGCTTGC  -1.98607349885094  0.137233215315936           mbnl1
          2   TGCTTGC  -2.09775518556902  0.259964844297944           mbnl1
          3   GGCTTGC  -2.38376620889964  0.352167512560293           mbnl1
          4   CGCTTGC  -2.49208691693447  0.434904633342322           mbnl1
          5   AGCTTGG  -3.67966981232818  0.460135937867991           mbnl1
          6   TGCTTGG  -3.79135149904627  0.482701022411463           mbnl1
          7   AGCTTAC  -3.9437269446445   0.502076889638399           mbnl1

> matt mk_uniq

Description:

Example:

         A    B    V1    V2
         a    1    1     2
         a    2    1     3
         b    2    1     4
         a    1    1     5

> matt mk_uniq t2.tab A B

         A    B    V1    V2
         a    1    1     2
         a    2    1     3
         b    2    1     4

> matt perm_seqs

Description:

> matt perm_seqs seqs.tab SEQ

> matt prnt_tab

Description:

Example:

t1.tab

> matt prnt_tab t1.tab -w 5 -s 3

> matt prnt_tab t1.tab -w 5 -s 3 | less -S

> matt retr_rnaseq

Description:

GEO

Example:

SRR3532927 nova2kd3
SRR3532926 nova2kd2
SRR3532925 nova2kd1
SRR3532924 nova2wt3
SRR3532923 nova2wt2
SRR3532922 nova2wt1

> matt retr_rnaseq accession_numbers.txt -keepsra -fasta -o rnaseq_data -p 6

dataset_info.tab
nova2kd1_1.fasta.gz
nova2kd1_2.fasta.gz
nova2kd2_1.fasta.gz
nova2kd2_2.fasta.gz
nova2kd3_1.fasta.gz
nova2kd3_2.fasta.gz
nova2wt1_1.fasta.gz
nova2wt1_2.fasta.gz
nova2wt2_1.fasta.gz
nova2wt2_2.fasta.gz
nova2wt3_1.fasta.gz
nova2wt3_2.fasta.gz
SRR3532922.sra
SRR3532923.sra
SRR3532924.sra
SRR3532925.sra
SRR3532926.sra
SRR3532927.sra

> matt sort_tab

Description:

Example:

t1.tab

> matt sort_tab t1.tab START

> matt sort_tab t1.tab START -d

> matt stratify

Description:

perm_test

Example:

DATASET: containing ndiff and down
GCC: containing the GC content of the exons

> matt stratify exons.tab GCC DATASET ndiff 0.05

col_calc

> matt cmpr_exons

Description:

run get_efeatures on the input table to determine exon-related features
apply Mann-Whitney U test to the comparison of feature distributions across groups
output:

PDF report summarizing results with box plots
table with all events and extracted features
table with details on performed statistical tests including p values
all box plots as PDF graphics

get_efeatures

start coordinate
end coordinate
chromosome ID (must match with chromosome IDs in GTF and FASTA; use extr_scafids to check IDs in GTFs and FASTAs)
strand
gene ID of gene where exon occurs in; these gene IDs must match with the gene IDs in the GTF which you use

Recommendation:

Example:

START
END
SCAFFOLD
STRAND
GENEID_ENSEMBL
DATASET: containing group IDs of exons (down, up, ndiff)

> matt cmpr_exons exons.tab START END SCAFFOLD STRAND GENEID_ENSEMBL ...

    ... Hsa19.gtf Hsa19.fa Hsap 150 DATASET[down,ndiff] down_vs_ndiff

Hint 1:

stratify

Hint 2:

Hint 3:

Hint 4:

> matt cmpr_introns

Description:

run get_ifeatures on the input table to determine exon-related features
apply Mann-Whitney U test to the comparison of feature distributions across groups
output:

PDF report summarizing results with box plots
table with all events and extracted features
table with details on performed statistical tests including p values
all box plots as PDF graphics

get_ifeatures

start coordinate
end coordinate
chromosome ID; these chromosome IDs must match with the chromosome IDs in the GTF and FASTA which you use
strand
gene ID of gene where intron occurs in; these gene IDs must match with the gene IDs in the GTF which you use

Recommendation:

Example:

START
END
SCAFFOLD
STRAND
GENEID_ENSEMBL
DATASET: containing group IDs of introns (down, up, ndiff)

> matt cmpr_introns introns.tab START END SCAFFOLD STRAND GENEID_ENSEMBL ...

    ... Hsa19.gtf Hsa19.fa Hsap 150 DATASET[down,ndiff] down_vs_ndiff

Hint 1:

stratify

Hint 2:

Hint 3:

Hint 4:

> matt rna_maps

Description:

Output:

PDF report containing all motif RNA-maps
PDF report with motif RNA-maps with significant enrichment only (with arguments -p, -fdr)
For each map, there are two versions, one with a data coverage section, and one without. The data coverage section shows the data overage along the map, which might be less then 100% if the data set contains exons/introns shorter than the specified exon/intron sections to be visualized.
The motif RNA-maps are ordered according to largest differences of motif enrichment scores; Maps with largest positive differences come first, maps with largest negative differences last. The reference group for computing differences is always the last group specified in the program call.
all motif RNA-maps as PDF graphics

Example:

TYPE: the type of motif model, either REGEXP or PWM
NAME: the name of the motif
REGEXP: the Perl regular expression of a link to a file (table) containing the PWM
THRESH: only for PWM (set to NA for REGEXPs): threshold for hit prediction (see get_pwm_hits)
BGMODEL: only for PWM (set to NA for REGEXPs): if and which background model should be used

START: start of exon
END: end of exon
SCAFFOLD: chromosome
STRAND
UPSTRM_EX_BORDER: border of next up-stream exon
DOSTRM_EX_BORDER: border of next down-stream exon
GROUP: defining exon groups (down, up, ndiff); last group is reference group

matt rna_maps exons.tab UPSTRM_EX_BORDER START END DOSTRM_EX_BORDER SCAFFOLD
    ... STRAND GROUP[up,down,ndiff] 31 35 135 Mmu09.fasta nova2_regexp.tab TYPE
    ... NAME REGEXP THRESH BGMODEL -d nova2_rnamap -p 0.0001 10000

Hint:

> matt rna_maps_cisbp

Description:

all ACGT k-mers (length 2-5)
all approx. 340 CISBP-RNA BPS with IUPAC binding motif
motif RNA-maps can be generated for exons or introns

Output:

PDF report containing all motif RNA-maps
PDF report with motif RNA-maps with significant enrichment only (with arguments -p, -fdr)
For each map, there are two versions, one with a data coverage section, and one without. The data coverage section shows the data overage along the map, which might be less then 100% if the data set contains exons/introns shorter than the specified exon/intron sections to be visualized.
The motif RNA-maps are ordered according to largest differences of motif enrichment scores; Maps with largest positive differences come first, maps with largest negative differences last. The reference group for computing differences is always the last group specified in the program call.
all motif RNA-maps as PDF graphics

START: start of exon
END: end of exon
SCAFFOLD: chromosome
STRAND
UPSTRM_EX_BORDER: border of next up-stream exon
DOSTRM_EX_BORDER: border of next down-stream exon
GROUP: defining exon groups (down, up, ndiff); last is reference group

matt rna_maps_cisbp exons.tab UPSTRM_EX_BORDER START END DOSTRM_EX_BORDER SCAFFOLD
    ...  STRAND GROUP[up,down,ndiff]  31 35 135 Hsa19.fasta cisbprna_regexps
    ...  -d cisbp_rnamaps

Hint:

rna_maps

> matt test_cisbp_enrich

Description:

CISBP-RNA

Example:

SEQ_3P: containing 3'-ends (150 nt) of introns
DATASET: containing group IDs of introns (e.g., down, up, ndiff)

> matt test_cisbp_enrich t.tab SEQ_3P DATASET[down,ndiff,up,ndiff]

get_pwm_prof

test_pwm_enrich

the entire sequences as given
only the first 1/3 part of the sequences
only the internal 1/3 part of the sequences
only the last 1/3 part of the sequences

Output:

SIGNIFICANT_RESULT_ALL: enrichment considering entire sequences
SIGNIFICANT_RESULT_FIRST: enrichment considering only the first 1/3 of the sequences
SIGNIFICANT_RESULT_INTERNAL: enrichment considering only the internal 1/3 of the sequences
SIGNIFICANT_RESULT_END: enrichment considering only the last 1/3 of the sequences

Number and length of sequences:

Avoid false positives:

cntr

Hint 1:

Hint 2:

the number of hits across sequences per group of sequences
the number of positive sequences containing hits per group

argument -m

Examples

Comparing sets of exons wrt. their exon features
Testing enrichment of TGC in up-stream regions of neural micro exons
Analyzing exons regulated by splicing factor Nova2

The examples contain many explanations and therefore appear lengthy. The more familiar the user is with Matt, the easier and faster the work experience will become.

Comparing sets of exons wrt. their exon features

data.tab

Irimia et. al, Cell, 2014

AS_noNeural
NEURAL-DOWN
NEURAL-UP

1. Making yourself familiar with data

> matt get_colnms data.tab

   ID    COL_NAMES
   1     GENE
   2     EVENT
   3     COORD
   4     LENGTH
   5     FullCO
   6     TYPE
   7     GROUP

> matt prnt_tab data.tab -W 30 | less -S -

   GENE EVENT         COORD                     LENGTH FullCO TYPE GROUP      
        HsaEX0000026  chr12:120648733-120648909 177    ...    S    AS_noNeural
        HsaEX0000032  chr12:10749491-10749580   90     ...    S    AS_noNeural
        HsaEX0000033  chr12:10748290-10748526   237    ...    C2   AS_noNeural
        HsaEX0000044  chr14:58760484-58760644   161    ...    C1   AS_noNeural
        HsaEX0000048  chr14:58752309-58752474   166    ...    C2   AS_noNeural
        HsaEX0000053  chr12:70636846-70636984   139    ...    S    AS_noNeural
        HsaEX0000062  chr14:101376570-101376641 72     ...    S    AS_noNeural
        HsaEX0000087  chr15:84850734-84850835   102    ...    S    AS_noNeural
        HsaEX0000091  chr15:57209263-57209410   148    ...    S    AS_noNeural
   A1BG HsaINT0000005 chr19:58858396-58858718   323    ...    IR-S AS_noNeural
   A2M  HsaINT0000025 chr12:9256997-9258831     1835   ...    IR-S AS_noNeural
   A2M  HsaINT0000026 cr12:9254271-9256834      2564   ...    IR-S AS_noNeural
   A2M  HsaINT0000027 chr12:9253804-9254042     239    ...    IR-S AS_noNeural

> matt col_uniq data.tab TYPE | matt prnt_tab - -w 9 | less -S -

   TYPE_UNIQ   FREQ     
   Alt3        3444     
   Alt5        2449     
   C1          1604     
   C1*         312      
   C2          1287     
   C2*         259      
   C3          1134     
   C3*         237      
   IR-C        3483     
   IR-S        15494
   ...

2. Adapting labels in column TYPE

get_vast

chg_labs

> matt chg_labs data.tab TYPE S*:S C1*:C1 C2*:C2 C3*:C3 MIC_M:MIC MIC_S:MIC > tmp.tab
> mv tmp.tab data.tab

3. Extracting exons and their start/end-coordinate, chromosome, strand

> matt get_vast data.tab COORD FullCO TYPE LENGTH | ...
   ... matt get_rows - TYPE]S,C1,C2,C3,MIC[ > exons.tab

> matt get_colnms exons.tab

   ID      COL_NAMES
   1       GENE
   2       EVENT
   3       COORD
   4       LENGTH
   5       FullCO
   6       TYPE
   7       GROUP
   8       SCAFFOLD
   9       START
   10      END
   11      STRAND
   12      UPSTRM_EX_BORDER
   13      DOSTRM_EX_BORDER

col_uniq

4.1 Adding Ensembl gene IDs

get_efeatures

cmpr_exons

get_match

mapping_EVENT2GENEID.tab

> matt get_match exons.tab EVENT mapping_EVENT2GENEID.tab EVENT ENSEMBL_GENEID ...
       ... | matt add_cols exons.tab -

4.2 Adding Ensembl gene IDs

retr_geneids

> matt retr_geneids exons.tab START END SCAFFOLD STRAND Hsa19_ensembl.gtf -f gene_id -n ENSEMBL_GENEID ...
       ... | matt add_cols exons.tab -

4.3 Adding Ensembl gene IDs

get_vast

> matt get_vast data.tab COORD FullCO TYPE LENGTH -gtf Hsa19_ensembl.gtf -f gene_id | ...
   ... matt get_rows - TYPE]S,C1,C2,C3,MIC[ | matt rn_cols - GENEID:ENSEMBL_GENEID > exons.tab

5. Checking number of exons in final table

> matt col_uniq exons.tab GROUP

   GROUP_UNIQ     FREQ
   AS_noNeural    10465
   NEURAL-DOWN    631
   NEURAL-UP      1046

6. Extracting exon features for all exons

sequences

gene annotation

twoBitToFa

> ./twoBitToFa -noMask hg19.2bit Hsa19_gDNA.fasta

> fasta=/..path..to../Hsa19_gDNA.fasta
> gtf=/..path..to../ensembl/Hsa19.gtf
> matt get_efeatures exons.tab START END SCAFFOLD STRAND ENSEMBL_GENEID ...
    ... $gtf $fasta Hsap 150 | matt add_cols exons.tab -

extr_seqs

7. Down-sample exon set AS_noNeural

rand_rows

> matt get_rows exons.tab GROUP]AS_noNeural[  ... 
       ... | matt rand_rows - 2000 132927352 > exons_testsets.tab
> matt get_rows exons.tab GROUP]NEURAL-DOWN,NEURAL-UP[ ...
  ... | matt add_rows exons_testsets.tab -

cmpr_exons

must not

> matt get_cols exons_testsets.tab START END SCAFFOLD STRAND ...
   ... ENSEMBL_GENEID GROUP > tmp.tab
> mv tmp.tab exons_testsets.tab

> matt col_uniq exons_testsets.tab GROUP

   GROUP_UNIQ     FREQ
   AS_noNeural    2000
   NEURAL-DOWN    631
   NEURAL-UP      1046

8. Statistically comparing exon sets and generating PDF report

> fasta=/..path..to../Hsa19_gDNA.fasta
> gtf=/..path..to../ensembl/Hsa19.gtf
> matt cmpr_exons exons_testsets.tab START END SCAFFOLD STRAND ENSEMBL_GENEID ...
   ... $gtf $fasta Hsap 150 GROUP[NEURAL-UP,NEURAL-DOWN,AS_noNeural] ...
   ... cmpr_1 -colors:brown2,darkgoldenrod2,azure4

PDF report

Overview of results of permutation tests across all comparisons and features
Box-plots for all numerical features with all exon groups tested
Comparative sequence plots for all groups of splice sites, branch points for all comparisons
Details of results of permutation tests

Testing enrichment of TGC in up-stream regions of neural micro exons

data.tab

AS_noNeural
NEURAL-DOWN
NEURAL-UP

Making yourself familiar with data
Adapting labels in column TYPE
Extracting exons and their start/end-coordinate, chromosome, strand

4. Down-sample exon set AS_noNeural

> matt col_uniq exons.tab GROUP

   GROUP_UNIQ   FREQ
   AS_noNeural  10465
   NEURAL-DOWN  631
   NEURAL-UP    1046

> matt get_rows exons.tab GROUP]AS_noNeural[ | ...
   ... matt rand_rows - 2000 923681819 > exons_testsets.tab
> matt get_rows exons.tab GROUP]NEURAL-DOWN,NEURAL-UP[ | ...
   ... matt add_rows exons_testsets.tab -

> matt col_uniq exons_testsets.tab GROUP

   GROUP_UNIQ   FREQ
   AS_noNeural  2000
   NEURAL-DOWN  631
   NEURAL-UP    1046

5. Retrieving up-stream sequences of exons

get_seqs

> fasta=/..path..to../Hsa19_gDNA.fasta
> matt get_seqs exons_testsets.tab START END 200 200 SCAFFOLD STRAND $fasta | ...
   ... matt add_cols exons_testsets.tab -

> matt get_colnms exons_testsets.tab

   ID  COL_NAMES
   1   START
   2   END
   3   SCAFFOLD
   4   STRAND
   5   TYPE
   6   LENGTH
   7   GROUP
   8   SEQ_UP
   9   SEQ_DOWN

> matt get_cols exons_testsets.tab SEQ_UP | less -S -

   SEQ_UP
   ...CAAGGCTGTCTTTTCTCCATAG
   ...TTAAATTGTCTTCTCTCCACAG
   ...ACTTGAATTTCTACTCCCACAG
   ...CAACTTTCTAATCTATCCCTAG
   ...CCCCCTTGTGTTCCTCCTGCAG
   ...ATTTTGTTATTTTTTTTTTTAG
   ...GCTGTTCGTTTTTCTTTCTTAG
   ...CTTCGCCTCCTTGCTGTTTCAG
   ...

> matt get_cols exons_testsets.tab SEQ_DOWN | less -S -

   SEQ_DOWN
   GTGAGCTGGCAGAAATGAAGGGACT
   GTAAGACCAGTTTACCTTTTGAAGA
   GTAATACTTTGGGGAAGGAAGTTTT
   GTAAGAGTATTTTCCTTAATCAGTT
   GTATGTGGCTGTGGTGAACGGGCAC
   GCATGTAGCACCACACCTAACTGGA
   GTAAGCCTGCCAGGAGCCACTCAGC
   GTGAGCTCTGAGCCTCTGGCATTTC
   ...

7. Defining sets of micro exons

nmic_s: neural, short micro exons with length 3-15
nmic: neural, micro exons with length 16-27
nex: neural exons with length > 27
ex: non-neural exons

add_val

> matt def_cats exons_testsets.tab GROUP2 ...
   ... 'nmic_s=LENGTH[3,15] GROUP]NEURAL-UP,NEURAL-DOWN[' ...
   ... 'nmic=LENGTH[16,27] GROUP]NEURAL-UP,NEURAL-DOWN[' ...
   ... 'nex=LENGTH[28,1000000000] GROUP]NEURAL-UP,NEURAL-DOWN[' ...
   ... 'ex=GROUP]AS_noNeural[' | matt add_cols exons_testsets.tab -

> matt col_uniq exons_testsets.tab GROUP2

   GROUP2_UNIQ	FREQ
   ex           2000
   nex          1371
   nmic         155
   nmic_s       151

7. Statistically testing enrichment of TGC in up-stream sequences

test_regexp_enrich

REGEXP   NAME
TGC      TGC

> matt test_regexp_enrich exons_testsets.tab SEQ_UP ...
   ... GROUP2[nmic_s,ex,nmic,ex,nex,ex] single tgc.tab NAME ...
   ... REGEXP quant 1000 > res.tab

res.tab

test_regexp_enrich

SIGNIFICANT_RESULT_ALL: enrichment considering entire sequences with results: nmic_s>ex ***, nmic>ex ***, nex>ex ***
SIGNIFICANT_RESULT_FIRST: enrichment considering only the first 1/3 of the sequences with results: none
SIGNIFICANT_RESULT_INTERNAL: enrichment considering only the internal/middle 1/3 of the sequences with results: nmic>ex ***, nex>ex ***
SIGNIFICANT_RESULT_END: enrichment considering only the last/end 1/3 of the sequences with results: nmic_s>ex ***, nmic>ex ***, nex>ex ***

8. Visualizing positional distributions of TGC in up-stream regions of exons

get_regexp_prof

> matt get_regexp_prof exons_testsets.tab SEQ_UP RL SEQ_DOWN LR tgc.tab ...
   ... GROUP2[nmic_s,nmic,nex,ex]

Matt: Positional distributions of hits of REGEXPs

Data file: exons_testsets.tab
Categories (number of sequences): ex ( 2000 ), nex ( 1371 ), nmic ( 155 ), nmic_s ( 151 )
Sequence columns selected (plotting parameter): SEQ_UP ( RL ), SEQ_DOWN ( LR )
K or file with information on REGEXPS: tgc.tab
Date: Mar 09, 2017

Ordered Results

Ordering from top to bottom in each column:

Reference group:

none

Densities all positions	Cumulative densities all positions	Densities first positions	Cumulative densities first positions
TGC	TGC	TGC	TGC

Plots

TGC - TGC

Back to Ordered Results

Downstream analysis of Nova2 regulated exons

Shell script

After having downloaded the RNA-seq data from GEO with help of Matt, and after having estimated inclusion levels of alternative-splicing events genome-wide with help of the program VAST-TOOLs, this downstream analysis consists of the following steps

Extraction of skipped exons from vast-tools output table, which contains skipped exons, retained introns, and alternative 3' and 5' splice sites.
Definition of three groups of skipped exons according to their delta inclusion levels (∆PSI) for Nova2 KD vs. WT: silenced (∆PSI≥25), enhanced (∆PSI≤-25), unregulated (-1≤∆PSI≤1).
Extraction of exon-related features, and statistical comparison of feature distributions with the Mann-Whitney U test for the detection of discriminative features.
The output encompasses:

a summary report (PDF) with box plots and p values
all box plots as PDF graphics
a table with extracted features for the exons
a table with details (test statistics, group sizes, p values) on performed statistical tests

Generation of motif RNA-maps for all approx. 340 RNA binding proteins with IUPAC binding motif from CISBP-RNA.
The output encompasses:

a summary report (PDF) with all motif RNA-maps. Each map is generated in two version, one containing a data coverage section and one which does not contain such a section.
all motif RNA-maps as PDF graphics

Generation of a specific motif RNA-map with a Nova2 binding motif derived from known Nova2 binding motifs, with highlighted regions of statistically significant (FDR≤0.01) different Nova2-motif enrichment.
The output encompasses:

a summary report (PDF) with all (in this case only one) motif RNA-maps. Each map is generated in two version, one containing a data coverage section and one which does not contain such a section.
all (in this case only one) motif RNA-maps as PDF graphics

Extraction of 50 nt long sequences upstream of exons
Statistical enrichment analysis in these extracted sequences with all CISBP-RNA RNA binding proteins with PWM binding motifs.
The output encompasses:

a table with detailed information on performed statistical tests (test statistics, p values, group sizes)
for all motifs their positional distribution along the 50 nt long sequences in all three groups as PDF and PNG graphics

Extending Matt: add new commands/functionality

new_command.pl

in sub-directory abilities

new_command.pl

other scripting/programming

FAQ / Trouble shooting

1. Which letters do genomic sequences contain?

test_cisbp_enrich

get_seqs

test_pwm_enrich

2. Which column names do my tables need to have?

get_seqs

3. I don't get a PDF report from cmpr_exons or cmpr_introns.

KOMA-Script
babel
graphicx
tocbibind

4. I get strange errors when extracting sequences from a FASTA.

How to cite

Splice site strengths

As Matt internally uses Yeo & Burge's maximum-entropy models for computing splice site strengths, please cite

Yeo et al., Maximum entropy modeling of short sequence motifs with applications to RNA splicing signals, 2003, DOI: 10.1089/1066527041410418

whenever you publish results related to splice site strength. This includes the following splice site features which you might have obtained with Matt's commands get_efeatures, get_ifeatures, cmpr_exons, cmpr_introns:
- MAXENTSCR_HSAMODEL_UPSTRM_5SS - maximum entropy score of 5ss of up-stream exon using a model trained with human splice sites
- MAXENTSCR_HSAMODEL_3SS - maximum entropy score of 3ss using a model trained with human splice sites
- MAXENTSCR_HSAMODEL_5SS - maximum entropy score of 5ss using a model trained with human splice sites
- MAXENTSCR_HSAMODEL_DOWNSTRM_3SS - maximum entropy score of 3ss of down-stream exon using a model trained with human splice sites

Binding motifs of RNA binding proteins from CISBP-RNA

As Matt uses binding motifs of RNA binding proteins as provided by the database CISBP-RNA, please cite

Ray et al., A compendium of RNA-binding motifs for decoding gene regulation, 2013, DOI: 10.1371/journal.pcbi.1001016

whenever you publish results obtained with Matt's command test_cisbp_enrich or when you have used any of the CISBP-RNA binding motifs in sub-folder data_external/cisbp_rna of Matt installation.

Predicted branch points

As Matt internally uses SVM_BPfinder for the prediction of branch points, please cite

Corvelo et al., Genome-Wide Association between Branch Point Properties and Alternative Splicing, 2010, DOI: 10.1038/nature12311

whenever you publish results related to predicted branch points. This includes the following features which you might have obtained with Matt's commands get_efeatures, get_ifeatures, cmpr_exons, cmpr_introns:
- DIST_FROM_MAXBP_TO_3SS_UPINTRON - distance to 3ss of best precited BP
- SEQ_MAXBP_UPINTRON - sequence of best predicted BP
- SCORE_FOR_MAXBP_SEQ_UPINTRON - BP sequence score of best predicted BP
- PYRIMIDINECONT_MAXBP_UPINTRON - Pyrimidine content between the BP adenine and the 3 prime splice site for best BP
- POLYPYRITRAC_OFFSET_MAXBP_UPINTRON - Polypyrimidine track offset relative to the BP adenine for best BP
- POLYPYRITRAC_LEN_MAXBP_UPINTRON - Polypyrimidine track length for best BP
- POLYPYRITRAC_SCORE_MAXBP_UPINTRON - Polypyrimidine track score for best BP
- BPSCORE_MAXBP_UPINTRON - SVM classification score of best BP
- NUM_PREDICTED_BPS_UPINTRON - number of all predicted BPs which have a positive BP score
- MEDIAN_DIST_FROM_BP_TO_3SS_UPINTRON - like DIST_FROM_MAXBP_TO_3SS but median over top-3 predicted BPs
- SEQ_BPS_UPINTRON - comma-separated list of sequences of top-3 predicted BPs
- MEDIAN_SCORE_FOR_BPSEQ_UPINTRON - like SCORE_FOR_MAXBP_SEQ but median over top-3 predicted BPs
- MEDIAN_PYRIMIDINECONT_UPINTRON - like PYRIMIDINECONT_MAXBP but median over top-3 predicted BPs
- MEDIAN_POLYPYRITRAC_OFFSET_UPINTRON - like POLYPYRITRAC_OFFSET_MAXBP but median over top-3 predicted BPs
- MEDIAN_POLYPYRITRAC_LEN_UPINTRON - like POLYPYRITRAC_LEN_MAXBP but median over top-3 predicted BPs
- MEDIAN_POLYPYRITRAC_SCORE_UPINTRON - like POLYPYRITRAC_SCORE_MAXBP but median over top-3 predicted BPs
- MEDIAN_BPSCORE_UPINTRON - like BPSCORE_MAXBP but median over top-3 predicted BPs
- DIST_FROM_MAXBP_TO_3SS_DOINTRON - distance to 3ss of best precited BP
- SEQ_MAXBP_DOINTRON - sequence of best predicted BP
- SCORE_FOR_MAXBP_SEQ_DOINTRON - BP sequence score of best predicted BP
- PYRIMIDINECONT_MAXBP_DOINTRON - Pyrimidine content between the BP adenine and the 3 prime splice site for best BP
- POLYPYRITRAC_OFFSET_MAXBP_DOINTRON - Polypyrimidine track offset relative to the BP adenine for best BP
- POLYPYRITRAC_LEN_MAXBP_DOINTRON - Polypyrimidine track length for best BP
- POLYPYRITRAC_SCORE_MAXBP_DOINTRON - Polypyrimidine track score for best BP
- BPSCORE_MAXBP_DOINTRON - SVM classification score of best BP
- NUM_PREDICTED_BPS_DOINTRON - number of all predicted BPs which have a positive BP score
- MEDIAN_DIST_FROM_BP_TO_3SS_DOINTRON - like DIST_FROM_MAXBP_TO_3SS but median over top-3 predicted BPs
- SEQ_BPS_DOINTRON - comma-separated list of sequences of top-3 predicted BPs
- MEDIAN_SCORE_FOR_BPSEQ_DOINTRON - like SCORE_FOR_MAXBP_SEQ but median over top-3 predicted BPs
- MEDIAN_PYRIMIDINECONT_DOINTRON - like PYRIMIDINECONT_MAXBP but median over top-3 predicted BPs
- MEDIAN_POLYPYRITRAC_OFFSET_DOINTRON - like POLYPYRITRAC_OFFSET_MAXBP but median over top-3 predicted BPs
- MEDIAN_POLYPYRITRAC_LEN_DOINTRON - like POLYPYRITRAC_LEN_MAXBP but median over top-3 predicted BPs
- MEDIAN_POLYPYRITRAC_SCORE_DOINTRON - like POLYPYRITRAC_SCORE_MAXBP but median over top-3 predicted BPs
- MEDIAN_BPSCORE_DOINTRON - like BPSCORE_MAXBP but median over top-3 predicted BPs

Predicted branch points

As SVM_BPfinder internally uses SVMlight, please cite

Joachims, Making large-Scale SVM Learning Practical. Advances in Kernel Methods - Support Vector Learning, B. Schölkopf and C. Burges and A. Smola (ed.), MIT-Press, 1999. (link)

whenever you need to cite SVM_BPfinder (point 3)

Binding score of human SF1

As Matt uses the human SF1 binding motif (position weight matrix) published by Corioni et al., please cite

Corioni et al., Analysis of in situ pre-mRNA targets of human splicing factor SF1 reveals a function in alternative splicing, 2011, DOI: 10.1093/nar/gkq1042

whenever you publish results related to this human SF1 binding motif. This includes the following features which you might have obtained with Matt's commands get_efeatures, get_ifeatures, cmpr_exons, cmpr_introns:
- SF1_HIGHESTSCORE_3SS_UPINTRON - highest score of a SF1 positon weight matrix trained with human data in the last 150 nt 3 prime intron positons of up-stream intron
- SF1_HIGHESTSCORE_3SS_DOINTRON - highest score of a SF1 positon weight matrix trained with human data in the last 150 nt 3 prime intron positons of down-stream intron

System requirements
•  Linux/Mac OS
•  Perl
•  R, Rscript
•  pdflatex
•  any PDF viewer
•  any Web browser

License
Matt is open-source and available
under GNU LGPLv3

Additional material
•  Intro to rna_maps

Matt paper
•  A. Gohr, M. Irimia,
   Matt: Unix tools for
   alternative splicing analysis,
   Bioinformatics, 2018, DOI:
10.1093/bioinformatics/bty606

Matt poster
•  At NGS 2017

Please cite as well
Matt uses great tools of others.

Contents

Introduction

Download and installation

Tables

Documentation of Matt commands

Matt: Positional distributions of hits of motif PWMs

Ordered Results

Plots

A2BP1 - a2bp1.tab

MBNL1 - mbnl1.tab

Matt: Positional distributions of hits of REGEXPs

Ordered Results

Plots

MBNL1 - [GC][CT]TT[GA]C

A2BP1 - [TA]GCATG[CA]|TACATA

Examples

Comparing sets of exons wrt. their exon features

Testing enrichment of TGC in up-stream regions of neural micro exons

Matt: Positional distributions of hits of REGEXPs

Ordered Results

Plots

TGC - TGC

Downstream analysis of Nova2 regulated exons

Extending Matt: add new commands/functionality

FAQ / Trouble shooting

How to cite