GO analysis followup

[gene_x]

Guys,
I have a specific question about Gene Ontology analysis. I have read a paper where they claim, say, immune response pathway is enriched in the top 500 most differentially expressed genes. I searched it and found its GO ID is GO:0006955. My question is where do you go and download all the genes and its annotations etc so I can easily process them?

I found a place for mouse, ftp://ftp.informatics.jax.org/pub/reports/index.html#go

and the gene_association.mgi file seems to be the one I'm looking for.

Could anyone share your experience with this if you have any?

Thanks!

[mudshark]

i usually fetch the latest annotations here: http://www.geneontology.org/GO.downl...otations.shtml

[gene_x]

Originally Posted by mudshark

i usually fetch the latest annotations here: http://www.geneontology.org/GO.downl...otations.shtml

I found the gene_association.mgi file in the link you mentioned as well. I downloaded it and it's almost the same as the one on MGI site.. slightly different.. I compared a few of the difference between two files and found the MGI site file has duplicated entries. Not sure how did that happen. Anyway, your site seems to be the one to go. Thanks!

[Richard Finney]

You can grab the flat files from NCBI and ungzip them ...
You'll be interested in gene2go and gene_info.

This script does the trick ...
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2accession.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2go.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2pubmed.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2refseq.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2sts.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2unigene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_history.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_info.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA...otkb_collab.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/go_process.xml
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2ensembl.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_group.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/mim2gene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2sts.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2unigene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2vega.gz

[gene_x]

Originally Posted by Richard Finney

You can grab the flat files from NCBI and ungzip them ...
You'll be interested in gene2go and gene_info.

This script does the trick ...
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2accession.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2go.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2pubmed.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2refseq.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2sts.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2unigene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_history.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_info.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA...otkb_collab.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/go_process.xml
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2ensembl.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_group.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/mim2gene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2sts.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2unigene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2vega.gz

Good to know this site. Thanks!

[gene_x]

Originally Posted by Richard Finney

You can grab the flat files from NCBI and ungzip them ...
You'll be interested in gene2go and gene_info.

This script does the trick ...
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2accession.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2go.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2pubmed.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2refseq.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2sts.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2unigene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_history.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_info.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA...otkb_collab.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/go_process.xml
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2ensembl.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_group.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/mim2gene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2sts.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2unigene.gz
wget -nc ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2vega.gz

These genes contain human, mouse and etc? It looks like so to me.. I guess it's much easier to use the GO annotation files instead of process two files and trying to put things together for one species.

[girlwithglasses]

Originally Posted by gene_x

I found the gene_association.mgi file in the link you mentioned as well. I downloaded it and it's almost the same as the one on MGI site.. slightly different.. I compared a few of the difference between two files and found the MGI site file has duplicated entries. Not sure how did that happen. Anyway, your site seems to be the one to go. Thanks!

It will be almost identical - the file on the GO website is a slightly filtered version of the one available at MGI.

Have you tried using AmiGO to download all the annotations to your GO term and its descendants? Search for the GO term at amigo.geneontology.org, view the annotations, and then filter so that you're only viewing annotations for your species of interest. You should be able to download the resulting annotation set.

[gene_x]

Originally Posted by girlwithglasses

It will be almost identical - the file on the GO website is a slightly filtered version of the one available at MGI.

Have you tried using AmiGO to download all the annotations to your GO term and its descendants? Search for the GO term at amigo.geneontology.org, view the annotations, and then filter so that you're only viewing annotations for your species of interest. You should be able to download the resulting annotation set.

I just tried that finally got the filter working and saw the output. I compared output to a file (grep those terms) and the output is basically the same, with different annotation date and slight difference in format.

I think it's easier to stick to one and I think the one from annotation download site is the best one in terms of consistency among term annotations.

[rnavon]

Why calculate the enrichment yourself instead of using an external tool?
GOrilla http://cbl-gorilla.cs.technion.ac.il/ for example can calculate the enriched GO terms for you.
It has 2 modes:
1. Rank all your genes (not just your differentially expressed ones) according to differential expression and GOrilla will find GO terms enriched at the top of your list.
2. Compare a target set (e.g. top 500 genes) to a background set (e.g. all your genes).

[gene_x]

Originally Posted by rnavon

Why calculate the enrichment yourself instead of using an external tool?
GOrilla http://cbl-gorilla.cs.technion.ac.il/ for example can calculate the enriched GO terms for you.
It has 2 modes:
1. Rank all your genes (not just your differentially expressed ones) according to differential expression and GOrilla will find GO terms enriched at the top of your list.
2. Compare a target set (e.g. top 500 genes) to a background set (e.g. all your genes).

Thanks for the note, GOrilla is a really catchy name

[mudshark]

Originally Posted by rnavon

2. Compare a target set (e.g. top 500 genes) to a background set (e.g. all your genes).

"ALL genes" is probably not a good background set. "ALL expressed genes" should be the universe.

[Richard Finney]

"ALL expressed genes" should be the universe.
Is that legit? I mean of course for multi-sample correction (FDR-Bonferroni)?

[rnavon]

Originally Posted by mudshark

"ALL genes" is probably not a good background set. "ALL expressed genes" should be the universe.

If a gene is not expressed in both groups. Doesn't it mean it is measured to be not differentially expressed between the two groups and should therefore be in the background set?

[dpryan]

Originally Posted by rnavon

If a gene is not expressed in both groups. Doesn't it mean it is measured to be not differentially expressed between the two groups and should therefore be in the background set?

Not necessarily, that will depend on biological replicate number, degree of expression in the group where it is found, how variance is calculated, and the particulars of the test being used for differential expression analysis.