I wonder if anyone could confirm whether some randomizing feature in MAQ's mapping algorithm could be responsible for different mappings of the same reads (with mapping quality over zero) in duplicate runs of the "map" command (with default setting).
I've seen in a few trials that duplicate runs (using the same bfq, bfa files) apparently produce a small number (~3% or so) of reads (after removing all that have a zero mapping quality) that map differently in the 2 executions. A tiny number map to different locations, slightly more map in only one execution (have a zero mapping quality in the other). The majority of the differences are in reads that map to the same location but get different mapping qualities, differing most often by only 1 or 2 points.
I confess my reading of the MAQ paper and documentation has not helped me decide if this is expected behavior.
Thanks.
I've seen in a few trials that duplicate runs (using the same bfq, bfa files) apparently produce a small number (~3% or so) of reads (after removing all that have a zero mapping quality) that map differently in the 2 executions. A tiny number map to different locations, slightly more map in only one execution (have a zero mapping quality in the other). The majority of the differences are in reads that map to the same location but get different mapping qualities, differing most often by only 1 or 2 points.
I confess my reading of the MAQ paper and documentation has not helped me decide if this is expected behavior.
Thanks.