YES: "Choice of transcripts and software has a large effect on variant annotation"

This post was inspired by Aaron Quinlan's tweet:

Poster 1485: variant annotation is harder than you think. transcript set matters (surprise). #ASHG2013 http://t.co/FgJzomqVuV

— Aaron Quinlan (@aaronquinlan) October 23, 2013

Here is an example of a missense mutation found with VCFPredictions, a simple tool I wrote for variant effect prediction.

#CHROM POS ID ALT REF 
1 23710805 rs605468 A G

my tool uses the UCSC knownGene track, here is the context of the variant in the UCSC genome browser. There is one exon for TCEA3 (uc021oig.1) on the reverse strand.


If the base at 23710805 is changed from 'A'→'G' on the forward strand, there will be a non-synonymous variation Y(TAT)→H(CAT) on the reverse strand.


At the NCBI rs605468 is said to be " located in the intron region of NM_003196.1."

VEP cannot find this missense variation:

Uploaded Variation Location Allele Gene Feature Feature type Consequence Position in cDNA Position in CDS Position in protein Amino acid change Codon change Co-located Variation Extra
rs605468 1:23710805 G - ENSR00001518296 Transcript regulatory_region_variant - - - - - rs605468 GMAF=A:0.1204
rs605468 1:23710805 G ENSG00000204219 ENST00000476978 Transcript intron_variant - - - - - rs605468 GMAF=A:0.1204
rs605468 1:23710805 G ENSG00000204219 ENST00000450454 Transcript intron_variant - - - - - rs605468 GMAF=A:0.1204

(of course, my tool doesn't find some variations found in VEP too)

That's it,
Pierre