blast2go mapping problem

[khopper]

I'm using blast2go to find GO mappings and annotations by importing protein sequences and their blastp hits, rather than running blastp within blast2go. However, only 7% of the proteins with blastp hits mapped to GO terms. This is very strange because I previously ran a version of this data set before I removed a very small number of genes that were bacterial contamination of the read libraries, and 40% of the proteins with blastp hits mapped to GO terms. I checked the formatting of the xml files from blastp for both data sets, and they appear identical. Has anyone encountered this problem? Is there a fix?

I tried the mapping again and got 36% maps for proteins with blastp hits. I have no idea why there was a difference between runs and to not think it worthwhile trying to figure out why.

[sgoetz]

The gene ontology mapping performance in Blast2GO depends and can vary mainly because of three things: The sequence itself, the blast result and the GO database used to do the mapping.

A good thing to look at is some of the blast and mapping statistics like species and e-value distribution for the blast results and in the mapping step the list of source-databases.

Also check a single blast results (context menu - show blast result ) and check the ACC/Symbol and Uniprot columns after the mapping. If no values or values other than gene ids or symbols appear in this columns there might have been a problem with the database you blasted against or during the parsing of the XML result files. In most cases this can be fixed adjusting some parsing parameters.

Also review the database configuration in the "File"-"Properties" menu to check to which Blast2GO GO Mapping DB version you had been connected to. Check that you are connected to the latest version.

Somethime sequences are also skipped during the mapping due to connection/network problems. You can can get results redoing these sequences. You can simply rerun the step since only non-mapped sequences will be analysed again. Already mapped/green sequences will not be send again which speeds up the thing and in case you had problems you can quickly check and/or recover some missing results.