SOAPsnp output

[Camg]

I was in the process of filtering the output file from soapsnp when I noticed something unusual about the sequencing depth. SOAPsnp tells me the seq depth of the best and second best base, as well as the overall seq depth at that site. I would expect that the addition of the number of best and second best bases at a particular site would add up to something close to (or exactly) the total sequencing depth. However, I'm seeing a few unusual cases like this:

chr03 82650 T W 0 A 35 127 127 T 35 124 124 1693 0 0.00000 1.00000 0

As you can see the total seq depth is 1693, but the addition of the number of best bases (A = 127) and second best bases (T = 124) is 251. If this were the case there would have to be 1442 C's or G's, which would make them more represented than the A's or T's and therefore they should be the best and second best bases.

Does anyone have any idea what's going on here? Has anyone seen this before?

I appreciate the help.

FYI: SOAPsnp output format:

1) Chromosome ID
2) Coordinate on chromosome, start from 1
3) Reference genotype
4) Consensus genotype
5) Quality score of consensus genotype
6) Best base
7) Average quality score of best base
8) Count of uniquely mapped best base
9) Count of all mapped best base
10) Second best bases
11) Average quality score of second best base
12) Count of uniquely mapped second best base
13) Count of all mapped second best base
14) Sequencing depth of the site
15) Rank sum test p_value
16) Average copy number of nearby region
17) Whether the site is a dbSNP.

[combiochem]

I was in the similar situation and found that my quality score offset is different from the default in soapsnp.
If you are using the sanger scheme for the quality score, you should put the option "-z !" when you run the soapsnp.