The PID here is so low that we won't be able to compare seqr. seqr's behavior will be different below ~40%; I actually have it set so that it only looks for similarity 50% and above.
This makes sense because PID below 40-50% is not really useful anyway.
The PID here is so low that we won't be able to compare seqr. seqr's behavior will be different below ~40%; I actually have it set so that it only looks for similarity 50% and above. This makes sense because PID below 40-50% is not really useful anyway.