We should consider reporting the fraction of distantly mapping reads, to only references of some minimum length. Poorly assembly will skew the count badly -- this is probably more likely in metagenomes.
Between metagenomes, differing assembly completeness will make these numbers had to compare.
Restriction to only refs > 10kb would probably be too severe. Something like 1kb or 500b.
We should consider reporting the fraction of distantly mapping reads, to only references of some minimum length. Poorly assembly will skew the count badly -- this is probably more likely in metagenomes.
Between metagenomes, differing assembly completeness will make these numbers had to compare.
Restriction to only refs > 10kb would probably be too severe. Something like 1kb or 500b.