Bloom Filter A error rate

[macmanes]

In your experience, what Bloom filter A error rate is achievable - 0.05, more, less? Should I attempt to optimize parameters such that error rate is minimized - if so, what parameters seem to be most influential.

[mourisl]

The error rate in my experience is around 0.01 when the sequencing error rate is about 1%, but can be higher if the sequencing error rate is high. And I believe 0.05 is still in the normal range. Moreover, Lighter takes the error rate of Bloom filter A into account, so 0.05 or even higher is fine. The output message is just for the ill-chosen sampling rate("alpha") or table size that is too small, which result in a Bloom filter error rate like more than 50%.