When testing low coverage (1x) assemblies with variable abundances, A5 can exit with a non-zero value causing the pipeline to halt.
If the same simulation is run with higher (more likely coverage), the non-zero exit goes away.
Ideally, we would offer some type of warning here, or perhaps we should revise the measure of coverage. As it happens, coverage multipliers are from the perspective of a metagenome where all references are equally abundant. For decaying abundance profiles with very low abundance genotypes, we can end up with dramatically fewer reads as coverage is simply treated as a multiplier.
When testing low coverage (1x) assemblies with variable abundances, A5 can exit with a non-zero value causing the pipeline to halt.
If the same simulation is run with higher (more likely coverage), the non-zero exit goes away.
Ideally, we would offer some type of warning here, or perhaps we should revise the measure of coverage. As it happens, coverage multipliers are from the perspective of a metagenome where all references are equally abundant. For decaying abundance profiles with very low abundance genotypes, we can end up with dramatically fewer reads as coverage is simply treated as a multiplier.