Bug: some reads are not placed in clusters properly according to pad/mincov

[ceiabreu]

Dear Mike,

We found that a small fraction of reads are not properly placed in clusters. I have a small reproducible example for you.

ShortStack --nohp --pad 1 --mincov 1 --readfile reads.fasta --genomefile genome.fasta

generates these wrongly split clusters (pasting from Counts.txt file): chr_1:17223-17246 Cluster_6 9 9 chr_1:17243-17263 Cluster_7 1 1

chr_2:333-355 Cluster_14 46 46 chr_2:356-377 Cluster_15 2 2

Hope this helps tracking it down.

Thanks,

Cei

ss_bug.zip

[MikeAxtell]

Thanks Cei, sorry to be so slow to respond. I am working through all of the open issues now, finally.

I have begun work on a major new release and I anticipate that the cluster-calling algorithm will be completely re-written. Which hopefully will solve whatever weirdness is going on here.

[MikeAxtell]

A large upgrade of ShortStack, version 4, is now in alpha testing. It is available if you want to test it on the ShortStack4 branch on GitHub. We are still doing some tests and optimizations; when it's ready we will merge to master and make a release. The new version has a re-written method for cluster calling that should fix this bug.