Mutect2 has two depth table "AS_SB_TABLE" vs "SB"

[gudeqing]

Instructions

• Hi, I am using gatk 4.3.0.0
• found a mutant filtered by "strand bias", please see the following record
• found two depth table named "AS_SB_TABLE" and "SB" in INFO column and FORMAT column respectively. Both table will result pvalue > 0.4, so why stand_bias reported ?
• The first sample is tumor sample ( the last second one in record line)

7	55259524	.	T	A	.	strand_bias	AS_FilterStatus=strand_bias;AS_SB_TABLE=1900,2286|20,18;DP=4360;ECNT=1;GERMQ=93;MBQ=20,20;MFRL=191,169;MMQ=60,60;MPOS=38;NALOD=3.13;NLOD=388.11;POPAF=6.00;ROQ=85;TLOD=51.42	GT:AD:AF:DP:F1R2:F2R1:FAD:SB	0/1:2536,38:0.015:2574:824,12:911,13:1809,26:1135,1401,20,18	0/0:1650,0:7.714e-04:1650:603,0:639,0:1291,0:765,885,0,0

Confusion

1. why there are two different tables for strand bias? which one should be used?

bug

1. with both tables, scipy.stats.fisher_exact([[ref_fwd, ref_bwd], [alt_fwd, alt_bwd]]) returns a pvalue over 0.4, so there should be no bias

Best wishes!