MAQ SNP calling

[bioinfosm]

> Hi,
>
> From my understanding MAQ is tested for 20-50x coverage regions for SNP
> calling. However, we routinely have projects where depth of coverage
> from Solexa reads is ~5000x.
>
> I see quite a few 'SNPs' that are not called by MAQ.
> These are the duplicate samples, where MAQ does NOT call a SNP
>
> Pos Depth %a %c %g %t
> 64 6632 0.11 92.01 0.69 7.19
> 64 6687 0.07 90.88 0.40 8.64
>
> While for these duplicates it makes a SNP call!
> 64 8594 0.05 91.91 0.54 7.49
> 64 7889 0.13 91.15 0.49 8.23
>
> Any suggestions on what threshold/filter to use to make consistent
> calling with high depth of coverage regions?

[bioinfosm]

SNP calling in high depth/coverage solexa data

Any other options to use apart from MAQ in calling SNPs on such data?

[apfejes]

Sort of on-topic....

I'm not sure what the answer is to your question about using MAQ, but I have had the pleasure of doing SNP calls on a MAQ .map file, for which I wrote my own SNP caller. Doing your own SNP calls is actually very easy, but leaves you with the same questions about filters (though, no consistency issues, that I'm aware of.).

[bioinfosm]

Yes, MAQ SNP calling is amazingly good! I just wish it scaled to the high-depth data I have. The other alternative which I will try today is take a random subset of my data and artificially get manageable depth

[Rao]

How Maq works for small indels?

[dvh]

How Maq works for small indels?

from the PDF manual/published paper: maq for indels needs paired end reads. maps read 1, then does gapped alignment of read 2. maq indelpe will list the indels called from the read2 gapped alignment.

try novocraft software if you want more indels from single end or paired end reads.

[Rao]

from the PDF manual/published paper: maq for indels needs paired end reads. maps read 1, then does gapped alignment of read 2. maq indelpe will list the indels called from the read2 gapped alignment.

try novocraft software if you want more indels from single end or paired end reads.

Thanks dvh,
How about SOAP

[bioinfosm]

I used SOAP to detect upto 4bp indels in the solexa data. I usually use it on the reads that MAQ is not able to align to reference.

[bioinfosm]

Another case of MAQ SNP calling that I would like to discuss here:

cns.final.snp
PKD1 37187 T C 255 255 1.12 63 62
PKD1 37189 T C 255 255 1.56 63 62

These are so close to each other, and look very likely to be False Positives. I was of the idea that MAQ throws such away, but not so..

[bioinfosm]

Missed SNP call; obvious from pileup output

I see this obvious SNP from the pileup file with 382 reads mapping the position of which 99.74% say its a C, as opposed to the reference T.

Still, the position is not reported in cns.snp file! As I understand, the maq cns2snp command on consensus.cns does not do any filtering and reports all SNPs. This ought to have been reported..

Anything I am missing here?

[phatjoe]

MAQ SNP calling doubts

Another case of MAQ SNP calling that I would like to discuss here:

cns.final.snp
PKD1 37187 T C 255 255 1.12 63 62
PKD1 37189 T C 255 255 1.56 63 62

These are so close to each other, and look very likely to be False Positives. I was of the idea that MAQ throws such away, but not so..

Hi Bioinfoism,

This is probably 4 years late, but have you resolve the SNP calling issue with MAQ? I am actually facing the same scenario here where neighbouring SNPs were called. Any ideas on what is causing this? Also, since neighbouring SNPs were called, can the alignment be trusted?

Rookie question. But thanks in advance