Query Regarding The Somatic Variation Called Using Varscan2

Question

[subhi.2k7]

I have used varscan to call somatic variations from tumor samples. I noticed one confusing thing. Varscan called a position as germline, even if it actually varied within the tumor and normal data of the sample. For example:

chrom    position    ref    var    normal_reads1    normal_reads2    normal_var_freq    normal_gt    tumor_reads1    tumor_reads2    tumor_var_freq    tumor_gt    somatic_status    variant_p_value    somatic_p_value    tumor_reads1_plus    tumor_reads1_minus    tumor_reads2_plus    tumor_reads2_minus                    
chr1    95348669    C    T    4    12    75%    T    12    28    70%    Y    Germline    1.05E-17    0.754378116    4    8    14    14    0    4    2    10
chr1    179417347    C    T    1    1917    99.95%    T    697    745    51.66%    Y    Germline    1    0.00E+00    348    349    371    374    1    0    1025    892
chr14    101529005    A    G    16    2690    99.41%    G    1002    1050    51.17%    R    Germline    1    0    669    333    699    351    12    4    1716    974
chr14    101532752    T    A    3    3373    99.91%    A    1539    1516    49.62%    W    Germline    1    0    825    714    812    704    2    1    1786    1587
chr15    100251015    C    G    1    2171    99.95%    G    856    901    51.28%    S    Germline    1    0    463    393    474    427    1    0    1094    1077

I also have one more question, what do the last 4 columns of the output from calling the somatic calls mean, as they do not have any headers?

I would be very obliged for any suggestions.

[Sean Davis]

The term "germline" means that the variant was present in the patient's germline (both normal and tumor) and, thus, not somatic. It does not mean that the locus is the same as the reference.

Answer 1

[Charles Warden]

For your main question, I think one advantage to using a somatic variant caller versus filtering the independent results is you can avoid marginal calls. So, if a variant barely meets the criteria in the tumor sample (say, it has 6 supporting reads when the minimum number of 5, whereas the tumor just misses the minimum threshold with say 3-4 supporting reads). Not sure if this applies to your specific case, but it is one possible reason.

Not sure about your second question - I was reviewing a VarScan somatic result just the other day and I know that number of headers exactly matched the number of values in the table (the tabs don't seem to be preserved when I copy over what you pasted in this post, otherwise I would have been more specific).

Either way, here is the legend for the column names:

http://varscan.sourceforge.net/somatic-calling.html#somatic-output