to decide whether a mutation is non-synonimous or not

[zslee]

i have short solexa sequences mapping to genome, in some positions there're
some certain enriched mutations, how do i know whether this mutation is non-synonimous or not ?
one method is download transcript isoforms and coding annotations from ucsc, then count from the coding start site to determine what codon this position belong to then go to codon - Aa table t ocheck, can anyone give me better method ? thanks in advance

also, i have checked, in biomart we can download phase information ~
maybe i will use this

ZSL

[brasj]

Is this human genomic? If it is, we have been using SIFT for this with great success.

[zslee]

Is this human genomic? If it is, we have been using SIFT for this with great success.

SIFT (Sorting Intolerant From Tolerant) is a program that predicts whether an amino acid substitution affects protein function
my work is simpler, i just want to know whether a change of bases on human genome affect the amino acide coded, maybe SIFT can be used next step of my project,
so thanks a lot ~~

[krobison]

Your original proposal is pretty much the definition of how to do it; under what metric is it less than ideal?

[swbarnes2]

I would think BLASTx is the tool of choice here. You determine where in your sequence your change is, and make sure that the BLASTx match covers that letter.

[krobison]

No, BLASTX is very poor choice for multiple reasons:

1) Misclassifications, both false positive & false negative, around splice junctions
2) Speed. Much slower than doing lookup.
3) Very short exons will be missed
4) More risk of confusing paralogs & pseudogenes with your actual gene

Mapping the position to annotated transcript(s) and then computing the new codon & comparing its translation to the old codon is fast, simple & guaranteed. Hard to see why one would choose slow & error-prone over that.

[brasj]

guys, give sift a look.

you stick all your variants in there and it outputs a list of all of them that are coding and the change it does in the protein. Additionally, it also gives them a score of "tolerability".

if I understood correctly, this does all you need... and more

Assuming this is human dna, of course.

[zslee]

thanks to you all
i 'll first use krobison's suggestion and consider sift later
~~

[colindaven]

There is another thread on this topic -

http://seqanswers.com/forums/showthread.php?t=3236

Myself and some others have already developed scripts for this.
Sift is good, but is for the human genome only.