MITAB I/O

Small library to work with, i.e. parse, PSI-MI TAB files (https://psicquic.github.io/).

This is NOT the official PSI-MI TAB library, neither pretends to replace it. To get the official Java client, try:

https://mvnrepository.com/artifact/psidev.psi.mi/psimitab

http://www.psidev.info/groups/molecular-interactions.

PSI-MI stands for Proteomics Standards Initiative - Molecular Interaction.

Long story short, this library helps to parse PSI-MI TAB files (like intact.txt.gz, or BIOGRID-ALL-3.5.185.mitab.zip).

Features

supports txt, gzip and zip files
reads from standard input
writes into standard output

There are 2 ways to use this library:

As a Java library
As a command line tool

As a Java library

Download source code

git clone https://github.com/pasculorente/mitab-io

Install library

mvn install

Import to your project

<dependency>
    <groupId>org.uichuimi</groupId>
    <artifactId>mitab-io</artifactId>
    <version>1.0-SNAPSHOT</version>
</dependency>

Reading

mitab-io provides one class to read PSI-MI TAB files (org.uichuimi.mitab.io.InteractionReader). This class allows you to iterate over the interactions of a file in 4 ways, choose your favourite:

read all
for loop
stream
iterator

import org.uichuimi.mitab.io.InteractionReader;
import org.uichuimi.mitab.io.model.Interaction;

import java.io.File;import java.util.List;

class Main { 
    public static void main(String[] args) throws Exception {
        File intact = new File("intact.txt.gz");
        try (InteractionReader reader = new InteractionReader(intact)) {
            // read all at once
            final List<Interaction> interactions = reader.readAll();

            // for loop
            for (Interaction interaction : reader) { 
            }

            // stream
            reader.interactions().filter(interaction -> true).forEach(interaction -> {});

            // as an iterator?
            while (reader.hasNext()) {
            	final Interaction interaction = reader.next();
            }
        } 
    }
}

Writing

Currently, we provide only 2 writers: tsv and neo4j.

Tab Separated Value (TSV)

Creates a TSV file with the following fields:

Remember that every PSI-MI field follows the syntax:

xref:value(description)|xref:value(description)...

For each column, we extract the first value

xref:value(description)|xref:value(description)...

Column	Name	PSI-MI column
1	ID	14
2	TYPE	12
3	METHOD	7
4	ORGANISM	29
5	SCORE	15
6	A_ID	1
7	B_ID	2
8	A_BIO_ROLE	17
9	B_BIO_ROLE	18
10	A_EXP_ROLE	19
11	B_EXP_ROLE	20

The file looks like

ID	TYPE	METHOD	ORGANISM	SCORE	A_ID	B_ID	A_BIO_ROLE	B_BIO_ROLE	A_EXP_ROLE	B_EXP_ROLE
EBI-7121552	MI:0407	MI:0084	-1	0.56	P49418	O43426	MI:0499	MI:0499	MI:0498	MI:0496
EBI-7121634	MI:0407	MI:0084	-1	0.44	P49418	EBI-7121639	MI:0499	MI:0499	MI:0498	MI:0496

Print to standard output

import org.uichuimi.mitab.io.InteractionReader;
import org.uichuimi.mitab.io.model.Interaction;

import java.io.File;import java.util.List;

class Main { 
    public static void main(String[] args) throws Exception {
        File intact = new File("intact.txt.gz");
        try (InteractionReader reader = new InteractionReader(intact);
            TsvWriter writer = new TsvWriter(System.out)) {
            reader.forEach(writer);
        } 
    }
}

Neo4j

neo4j-admin import and cypher IMPORT CSV use a particular TSV/CSV file format. The ways a PSI-TAB file can be split into edge and node files are many, and the discussion of which one is best is out of this scope. We present a way of splitting them, which will serve for you to implement your own writer.

import org.uichuimi.mitab.io.InteractionReader;
import org.uichuimi.mitab.io.model.Interaction;

import java.io.File;import java.util.List;

class Main { 
    public static void main(String[] args) throws Exception {
        File intact = new File("intact.txt.gz");
        try (InteractionReader reader = new InteractionReader(intact);
            TsvWriter writer = new TsvWriter(System.out)) {
            reader.forEach(writer);
        } 
    }
}

Created files are:

nodes.tsv.gz

Column	Name	PSI-MI column
1	identifier:ID(interactor)	1 / 2

Example:

identifier:ID(interactor)
EBI-7121639
EBI-7121654
EBI-7121715
EBI-7121765
EBI-7121785
EBI-7121821

relationships.tsv.gz

Column	Name	PSI-MI column
1	:START_ID(interactor)	1
2	:END_ID(interactor)	2
3	identifier	14
4	type	12
5	method	7
6	score	15

Example:

:START_ID(interactor)	:END_ID(interactor)	identifier	type	method	score
P34709	P34708	EBI-640637	MI:0915	MI:0047	0.51
P34708-1	P34709	EBI-367231	MI:0915	MI:0018	0.51
P49418	EBI-7121639	EBI-7121634	MI:0407	MI:0084	0.44
P49418	EBI-7121654	EBI-7121659	MI:0407	MI:0084	0.44
P49418	EBI-7121715	EBI-7121710	MI:0407	MI:0084	0.44
EBI-7121765	P49418	EBI-7121760	MI:0407	MI:0084	0.44

As a command line tool

NOTE: this option is the less recommended, since this script provides little functionality.

Compile and package the library into a jar.

mvn clean compile assembly:single

Check if it works

java -jar target/mitab-io-1.0-SNAPSHOT-jar-with-dependencies.jar --help

You should see something like

Usage: mitab [-hv] [-i=<input>] [--neo4j=<neo4j>] [-o=<output>]
umpteenth package with tools to work with PSI MITAB files
  -h, --help              display this help message
  -i, --input=<input>
      --neo4j=<neo4j>
  -o, --output=<output>
  -v, --verbose

With this option you can export PSI-MI TAB files into tsv or neo4j-like files.

java -jar target/mitab-io-1.0-SNAPSHOT-jar-with-dependencies.jar \
    --input my-file.txt.gz \
    --neo4j neo4jfiles \
    --output simplified.tsv