Comparing very similar sequences does not provide all results.

[KasperThystrup]

First of thanks for a great tool!

While playing around with some comparisons between genes from the same file, I noticed that blastn behaves differently from expected:
blastn -query genes.fasta -subject genes.fasta -outfmt 6

Having two identical genes (cps2B & cps8B) in genes.fasta - results in following matches:
cps2B:cps8B (100% id and cov)
cps8B:cps2B (100% id and cov)

This comparisson misses the following:
cps2B:cps2B
cps8B:cps8B

Now adding a third gene (cps7B) to the mix by appending it to the genes.fasta, changes everything out entirely:
cps2B:cps2B
cps2B:cps7B (99.76% ID and 100% cov)
cps8B:cps8B
cps7B:cps7B
cps7B:cps2B (99.76% ID and 100% cov)

This comparisson misses the following:
cps2B:cps8B
cps8B:cps2B
cps8B:cps7B (should be 99.76% as well as cps2B and cps8B are identical)
cps7B:cps8B (should be 99.76% as well as cps2B and cps8B are identical)

Is there a way to include all top matches?

[JacobLondon]

Glad you enjoy the tool! It's been a while since I've looked at my undergrad senior project. A quick disclaimer, I want to note that I have very little experience in bioinformatics other than this project, although our professor Mohamed El-Hadedy Aly, Ph.D. at California Polytechnic University, Pomona could give a very educated answer to possible questions.

That out of the way, I looked at the matching method in extend.cpp where the implementation attempts to find the best match via scoring/extending with the Smith-Waterman algorithm. It could be feasible to modify the extend_filter member function from 'track the best at all times' approach to keeping a std::List of ExtendedSequenceMap objects that is then sorted by score and returned, providing top matches.

Unfortunately, I haven't the time to maintain this project, but if you were so technically inclined, send me a pull request with an implementation and I might be able to provide that as an alternate approach!