go_enrichment

Transcripts annotation and GO enrichment Fisher tests

Overview

go_enrichment annotates transcript sequences and performs GO enrichment Fisher tests. The transcript sequences are blasted against the swissprot protein database and the uniprot information corresponding to the hit is retrieved from the uniprot website. Fisher tests are performed with the goatools Python module.

Prerequisites

To use go_enrichment, you will need a UNIX system (Linux or Mac OSX) and the following dependencies installed on your computer (see Installation section for more details about installing these prerequisites):

• wget
• gnu parallel
• blastplus 2.7.1+, the NCBI suite of blast tools
• swissprot blast database ftp://ftp.ncbi.nlm.nih.gov/blast/db/swissprot.tar.gz
• GO database (see GO database section below)
• goatools

Installation

If you do not have administrator rights on the computer you will be using or have little experience compiling, installing and adding programs to your PATH environment variable, you will potentially need to ask an administrator to install the following programs and databases.

Wget

Your UNIX system should already have wget installed. Test this by running:

wget

If you get a message saying there is a missing URL, wget is installed. Otherwise, if you are using a computer with OSX. Google install wget OSX and follow the installations instructions. For Debian or Ubuntu derived Linux distributions, install wget with:

sudo apt-get install wget

Gnu Parallel

We will use wget to download gnu parallel:

wget http://ftp.gnu.org/gnu/parallel/parallel-latest.tar.bz2
tar xvfB parallel-latest.tar.bz2
cd parallel-*
./configure && make && sudo make install

Blast tools 2.7.1+

The blast executables (pre-compiled for different architectures) can be found here: ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/. Download the right ones for your computer, uncompress the archive and copy all the files that are in the bin folder so that they are accessible through the $PATH variable on your system.

For example, if you have administrator rights on the system, you could do:

sudo cp /path_to_blastplus/bin/\* /usr/local/bin

Test the installation by launching blastn's help:

blastn -h

Swissprot database

We will use wget to download the swissprot databases.

# Create a temporary bash session
bash

# Create folder to contain the databases
mkdir ~/blastplus_databases
cd ~/blastplus_databases

# Downloading the database
wget ftp://ftp.ncbi.nlm.nih.gov/blast/db/swissprot.*

# Confirming the integrity of the downloaded files
cat *.md5 | md5sum -c

# Decompressing
for file in `ls swissprot.*.gz`; do tar -xzf $file ; done

# Exit temporary bash session
exit

GO database

Installing the GO database will be faster:

# Create a temporary bash session
bash

# Moving to the GO database folder
cd 02_go_database

# Downloading the GO databases
wget http://geneontology.org/ontology/go-basic.obo

# Exit temporary bash session
exit

goatools

NEW WAY TO INSTALL (2021-06-30)

conda create -n goatools -c bioconda goatools=1.1.7
conda activate goatools

Old way

goatools is a python module. It depends on a certain number of other python modules. In order to make the installation easier, we will be using the anaconda python data analysis platform. anaconda will make it easy to install most of the module dependencies and does not require administrator rights. To get the anaconda install file, go to https://www.continuum.io/downloads and choose the appropriate platform and python 2.7, then launch the installation and follow the instructions. When asked if you want anaconda to add itself to your $PATH variable, say yes. You can then update with:

conda update conda

Then go to the goatools GitHub page https://github.com/tanghaibao/goatools and follow the installation instructions.

Workflow

This is a brief description of the steps as well as the input and output formats expected by go_enrichment.

Step 1 - Blast against swissprot

Put your sequences of interest in the 03_sequences folder in a file named transcriptome.fasta. If you use another name, you will need to modify the SEQUENCE_FILE variable in the script.

You need the script to point to the locally installed blastplus database by modifying the SWISSPROT_DB variable.

Then run:

./01_scripts/01_blast_against_swissprot.sh

Step 2 - Get annotation information from uniprot

This step will use the blast results to download the information of the genes to which the transcript sequences correspond.

Run:

./01_scripts/02_get_uniprot_info.sh

Step 3 - Annotate the transcripts

Use this step to create a .csv file containing the transcript names as well as some annotation information (Name, Accession, Fullname, Altnames, GO).

Run:

./01_scripts/03_annotate_genes.py 03_sequences/transcriptome.fasta 05_annotations/ sequence_annotation.txt

Step 4 - Extract genes

Before we can perform the Fisher tests, we need to generate two text files containing (one per line):

• The names of all the analyzed transcripts, 'all_ids.txt'
• The names of the significant transcripts, 'significant_ids.txt'

Step 5 - Run goatools

WARNING! This is currently broken. Follow the next steps to use goatools:

Install goatools

See Installation section, including getting the GO databases https://github.com/tanghaibao/goatools

Run goatools

python2 scripts/find_enrichment.py --pval=0.05 --indent ../wanted_transcripts.ids ../all_ids.txt ../all_go_annotations.csv > ../go_annotation.tsv

This script will launch goatools and perform the Fisher tests. Note: edit the script to point to your own installation of find_enrichment.py

TODO put back in the following script

./01_scripts/04_goatools.sh

Step 6 - Filter goatools results

WARNING!

This script no longer works with goatools 1.1.6. Better to filter by hand.

We can now reformat the results of goatools to make them more useful.

./01_scripts/05_filter_goatools.py enrichment.csv 02_go_database/go-basic.obo filtered.csv

Licence

go_enrichment is licensed under the GPL3 license. See the LICENCE file for more details.