Distribution of GO annotations among human genes

Recently, someone emailed me to ask how I got the data behind this figure that I often use in the introductions of my talks:

5-24-2014 4-54-08 PM

This figure shows that while there are few genes that are very well annotated (100s of GO annotations manually annotated by biocurators), that degree of annotation falls off very quickly. At the time I last did the analysis, 65% of genes had 5 or fewer GO annotations, and 41% had one or zero GO annotations.

To generate the image, I processed the gene2go.gz file at ftp://ftp.ncbi.nih.gov/gene/DATA. The one-liner to do the analysis is here:

wget -qO- "ftp://ftp.ncbi.nih.gov/gene/DATA/gene2go.gz" | gzip -d | awk '$1==9606&&$4!="IEA"{print $2}' | sort | uniq -c | sort -k1nr | less

The first line is the count of GO annotations, and the second column is the Entrez Gene ID. The first ten lines are:

149 1499
125 7422
123 23411
121 2263
115 7040
109 7157
108 652
106 4208
105 153090
105 348

So the first line says that beta-catenin (with Entrez Gene ID: 1499) had 149 unique GO annotations.

Posting this here in case someone else finds that useful! (And yes, I am a snob about one-liners. awk, sed, sort, uniq, join and tr are every bioinformatician’s friend!)


Leave a Reply

Your email address will not be published. Required fields are marked *