Potential issue in 'Calling variants in non-diploid systems'

In the section 'Calling non-diploid variants > Filtering variants' I think the suggested filters for SRP, SAP and EPP are incorrect and should be 'SAP < 20' etc (i.e. select for low biases rather than high values). These metrics quantify the probability of observing the strand distributions assuming random allocations between e.g. forward and reverse strands, so a high phred-score indicates the observed distribution is extreme/highly unlikely (therefore indicating a strand bias).

The selected variants after filtering show high strand bias (e.g. chrM pos 3243 variant: SAF = 443, SAR = 922, with a high SAP = 368.01) whereas some of the variants removed have more balanced strand distributions (e.g. chrM pos 263 variant: SAF = 172, SAR = 192, with a lower bias SAP = 5.39). Perhaps the interpretation of these scores could be clarified in this section @nekrut @astrovsky01?

Start in the training-material section “Calling non-diploid variants > Filtering variants” and compare the documented SRP, SAP, and EPP filters with the metric definitions. Use the chrM examples in the issue to verify the interpretation; done means the filters select low bias scores and the explanation matches the corrected behavior.

Written by the indexing model from the issue text.