Adaptive HSP threshold with LAV produces erroneous output

[rsharris]

When an adaptive HSP threshold is used, forward-strand alignments in LAV output are not properly tagged as forward strand. Downstream LAV readers will place these alignments on the reverse strand.

The HSP threshold is adaptive if --hspthresh=top<basecount> or --hspthresh=top<percentage>% is used.

LAV is the default output format. This issue is not observed in any other output formats.

Specifically, in this condition lastz fails to emit the s-stanza for the forward strand. Instead, it emits the s-stanza for the reverse strand, then the reverse strand a-stanzas, then the forward stand a-stanzas.

In typical use, lastz processes the forward strand first, independent of the reverse strand, and emits all forward strand alignment info before proceeding with the reverse strand. In contrast, when an adaptive HSP threshold is used output of forward alignments has to be delayed until after reverse strand alignments are found -- which alignments are output depends on what is found on both strands. The bug resulted from the fact that when the forward strand alignments were finally output, the need to output any strand-related pre-alignment information was neglected.

A correction for this will be made shortly. A new tagged release will follow soon after, incorporating a few much more minor fixes.

[rsharris]

The new tagged release 1.04.03 contains the fix for this.