What steps will reproduce the problem?
1. MosaikAligner -in test.CS.dat -out test.aligned.dat -ia ref.CS.dat -ibs
ref.dat -mm 6 -hs 15 -mhp 100 -act 20 -j h.sapiens_15
2. MosaikText -in test.aligned.dat -sam test.aligned.sam -u
3.
What is the expected output? What do you see instead?
When visualizing the alignment in IGV several clearly weird misalignments
appear. See the attached file for examples of the weird things.
What version of the product are you using? On what operating system?
MosaikAligner 1.0.1388 (also tried on MosaikAligner 1.0.1384).
On 2.6.31-gentoo-r6, 64-bit
Please provide any additional information below.
Most of the alignments are as expected but there are a bunch of this weird
reads included. Almost all of this weird alignments have alignment scores
above 20 and some even have as high as above 40.
In the first attachment Mosaikproblem.doc the aligner chooses to do an
insertion as base 5 giving a base shift with lots of mismatches as a
result. Doing a mutation at base 5 and a deletion in the end would give an
almost perfect alignment. There are 2 questions, 1) Why is it better to do
an insertion instead of mutation when the insertion gives a mismatches for
the rest of the read, 2) Why are the badly aligned sequence kept)?
In Mosaikproblem2.doc & Mosaikproblem3.doc only a few bases in the
beginning matches and rest seems to be something close to nonsense. Why are
these alignements kept? Shouldn't they be filtered away with mm set to 6?
Original issue reported on code.google.com by selekto...@gmail.com on 1 Mar 2010 at 10:57
Original issue reported on code.google.com by
selekto...@gmail.comon 1 Mar 2010 at 10:57Attachments: