How To Retrieve A Gene If Its Name Has Changed In A Gds?

Question

[Olivier]

Hello

I have a gene expression dataset file where I expected to see a particular gene in there, as per the published paper, but couldn't find it. Is there any way of retrieving the new name (even probe id, if possible?) of the gene?

In any case, do you think the annotations might change drastically so as to assign completely new functions to the same genes? Does it happen often? Are there any shortcomings as a result of that?

Thanks

[Neilfws]

Can you link to the GDS and tell us the gene that you were expecting to find?

[Olivier]

TRPV1 from GDS2642. I've posted my steps in the "bioinformatics study group". There's also code from other people. Well it's a beginning (Bioinformatics study group)
zcat GDS2642_full.soft.gz|grep 'TRPV1'
Link to dataset: ftp://ftp.ncbi.nlm.nih.gov/geo/datasets/GDS2nnn/GDS2642/soft/GDS2642_full.soft.gz
paper used: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3479650/

Answer 1

[Neilfws]

My first approach: use BioMart to map HGNC gene symbol to U95AV2 probe:

Ensembl Gene ID    Ensembl Transcript ID    Affy HG U95AV2 probeset    EntrezGene ID    HGNC symbol
ENSG00000196689    ENST00000399759                                     7442             TRPV1
ENSG00000196689    ENST00000571088                                     7442             TRPV1
ENSG00000196689    ENST00000399756                                     7442             TRPV1
ENSG00000196689    ENST00000425167                                     7442             TRPV1
ENSG00000196689    ENST00000574085                                     7442             TRPV1
ENSG00000196689    ENST00000576351                                     7442             TRPV1
ENSG00000196689    ENST00000310522                                     7442             TRPV1
ENSG00000196689    ENST00000570742                                     7442             TRPV1
ENSG00000196689    ENST00000174621                                     7442             TRPV1

Nothing. That's interesting. A quick look at the UCSC genome browser. Again, nothing.

Conclusion - the current annotation of U95AV2 does not map to TRPV1. Why? Speculation - U95 is a very old Affymetrix platform. It may be that with improved annotation of both probeset and gene location over time, features have shifted such that they no longer align.

Possible solution? Visit the Affymetrix product page, download the probe sequences in FASTA format for HG_U95Av2 (last updated 2008), align them to the current genome build using e.g. Bowtie 2, see which (if any) probeset(s) overlap TRPV1.

However, I note that in the publication, additional evidence for TRPV1 comes from qPCR, so the microarray data may be a moot point anyway.

I would add: issues like this with GEO datasets make me highly-sceptical that efforts to mine multiple experiments from GEO are informative.