Interpretation of columns in MAQ SNP output?

[griffon42]

Hi all-

Apologies if this information has been posted elsewhere - I wasn't able to find what I was looking for searching the forums.

I'm using MAQ cns2SNP (and SNPfilter) to call SNPs on short read alignments, mouse genome. Though things have been going well, I'm having a hard time interpreting some of the output.

The MAQ manual states, for cns2SNP output:
"Each line consists of chromosome, position, reference base, consensus base, Phred-like consensus quality, read depth, the average number of hits of reads covering this position, the highest mapping quality of the reads covering the position, the minimum consensus quality in the 3bp flanking regions at each side of the site (6bp in total), the second best call, log likelihood ratio of the second best and the third best call, and the third best call. "

I'm having trouble interpreting "the average number of hits of reads covering this position." Can anyone translate this into something more well-defined?

Also, for "Phred-like consensus quality" - though I realize this is not a true Phred score, what is the best way to think about this? Overall consensus base quality at a given position? "Confidence score" for calling a SNP?
In general, using SNPfilter and some arbitrary filters (discussed on the forums), I've been ignoring called SNPs with a "Phred-like" score of <40. I'd like to know what this actually means and whether it makes good sense.

Thanks for you help!

[der_eiskern]

re: Griffon interpreting SNPfilter outputs

Hi all-

Apologies if this information has been posted elsewhere - I wasn't able to find what I was looking for searching the forums.

I'm using MAQ cns2SNP (and SNPfilter) to call SNPs on short read alignments, mouse genome. Though things have been going well, I'm having a hard time interpreting some of the output.

The MAQ manual states, for cns2SNP output:
"Each line consists of chromosome, position, reference base, consensus base, Phred-like consensus quality, read depth, the average number of hits of reads covering this position, the highest mapping quality of the reads covering the position, the minimum consensus quality in the 3bp flanking regions at each side of the site (6bp in total), the second best call, log likelihood ratio of the second best and the third best call, and the third best call. "

I'm having trouble interpreting "the average number of hits of reads covering this position." Can anyone translate this into something more well-defined?
...
In general, using SNPfilter and some arbitrary filters (discussed on the forums), I've been ignoring called SNPs with a "Phred-like" score of <40. I'd like to know what this actually means and whether it makes good sense.

What I can gather is that the "average number of hits of reads covering this position" refers to the repetitiveness of the reads mapped to that locus. My impression was that many reads will map to more than 1 genomic position. Anything greater than ~1.1 I consider repetitive and ignore (if i can). This was suggested by Shen et al. According to them 1.1 "represents a conservative cutoff to avoid repeats and alleviate the mapping issues with shorter...reads."

as for phred-like quality scores, i believe heng li talks about what that means in section 4.5 of the first MAQ paper: http://genome.cshlp.org/content/18/11/1851.full.
quickly, (mathematics omitted)

Before consensus calling, MAQ first combines mapping quality and base quality. If a read is incorrectly mapped, any sequence differences inferred from the read cannot be reliable. Therefore, the base quality used in SNP calling cannot exceed the mapping quality of the read. MAQ reassigns the quality of each base as the smaller value between the read mapping quality and the raw sequencing base quality.

hope this helps.

[griffon42]

Thanks. That makes things considerably more clear.

[mixter]

Allele counts from MAQ snp output

Hello,

I have an additional question, after having read this thread and the documentation.

What I need would be the direct counts for the reference and variant allele, like in VarScan, for example, but in MAQ output there is just "read depth"?

Is there any way I can get or derive these 2 counts (reference and variant) from MAQ SNP output?

Thanks!