./VarScan.v2.3.6.jar mpileup2snp /home/data/SNP/archive/Novo_H15_q20.mpileup --min-coverage 8 --min-var-freq 0.01 --min-avg-qual 20 --p-value 99e-02 --min-reads2 3 --strand-filter 1 > /home/data/SNP/archive/Novo_H15_q20_Var_cov8_var0.01_Q0_p99e02_MinR3_no_strand.out Only SNPs will be reported Min coverage: 8 Min reads2: 3 Min var freq: 0.01 Min avg qual: 20 P-value thresh: 0.99 Reading input from /home/data/SNP/archive/Novo_H15_q20.mpileup
There are some SNPs being called which are not SNP's. For instance with three samples the coverage is incorrect 50:46:36 when look at bam files and there is the same mutation in each of the samples at the same position as similar frequencies however the SNP is only shown in only being detected in one of the samples (I've taken into account SNP quality and strand bias):
sample1 sample2 sample3 coverage A1 12690 A1 C T 25 0 0 16:14:23 N
Original out from varscan looks like this:
Chrom Position Ref Var Cons:Cov:Reads1:Reads2:Freq:P-value StrandFilter:R1+:R1-:R2+:R2-:pval SamplesRef SamplesHet SamplesHom SamplesNC Cons:Cov:Reads1:Reads2:Freq:P-value A1 C T Y:53:49:4:7.55%:9.8E-1 Pass:0.0:8:4:0:4:3.8462E-2 2 1 0 0 Y:16:12:4:25%:9.8E-1 C:14:14:0:0%:1E0 C:23:23:0:0%:1E0
I do get warnings when running varscan such as: Warning: error generating consensus from DEL-1-
Maybe this has an effect? Can anyone see why the SNP is not being called in the other two samples and thus resulting in a program induced false positive. Varscan seems a bit buggy, if I don't set the p-value at the command line the default value is not what it should be.