Problems with LCA quorum about sina HOT 7 OPEN

I'll have a look at it. Note on the side, you can write -vvv instead of -v -v -v. I assume you were using the 1.7.0?

from sina.

Thanks :) Yes, I used 1.7.0 but also tried 1.6.1 and 1.2.11 and the classification result was the same with all versions (did not check the nearest neighbors in details with the old versions, though).

from sina.

mind if I add the example to the unit tests?

from sina.

Simple enough. Classic int/float truncation issue. Should work now. I'm pushing 1.7.1, should be in bioconda within a day or so.

from sina.

mind if I add the example to the unit tests?

No, go ahead.

The submitted fix only fixes the difference between the second and third execution. The problem that the first test case

./sina -i aligned.query.fasta --prealigned --fs-kmer-no-fast --search --search-db SILVA-no-fields-reduced.opt.arb --search-kmer-len 10 --search-min-sim 0.95 --lca-quorum 0.8 --search-no-fast --lca-fields tax_slv -vvv

returns Archaea;Halobacterota; instead of Archaea;Halobacterota;Methanomicrobia;Methanomicrobiales;Methanomicrobiaceae;Methanoculleus; despite 8 of the 10 top hits having the latter classification is still present in 1.7.1.

from sina.

Darn - didn't see that part. Case of TL'DR I suppose.

from sina.

You are correct in your observation. The algorithm does an LCA between the best match and the others. Guess that was usually so close to what it's supposed to do that no-one notices. This will take me a day or two to fix. Don't have enough time in the evenings.

from sina.