Does it make sense to repeat the assembly step in variant calling?

Question

[ajuiwl]

Since the mapping step is non-deterministic, would it make sense to generate another BAM file with the same FASTQ file and same config with the objective of analyzing if a new variant is detected?

Thank you very much

Answer 1

[Devon Ryan]

No, since the results are at the very least "mostly deterministic" (I assume you're using bowtie2, which is the only aligner I know of that mentions this). It's vastly more efficient to simply filter with a reasonable MAPQ threshold.

[ajuiwl]

Thank you for your answer!!