Closed vpbrendel closed 5 months ago
Hi @vpbrendel,
Thank you for reporting this issue. I also noticed this recently and already made a fix for it. The issue happens exclusively when working on small input data resulting in a very low number of minimizers and/or k-mers in the graph. This currently causes a rounding issue in the MinimizerIndex and KmerHashTable when trying to reallocate these tables to larger sizes which in turns causes the program to hang as it indefinitely tries and fails to reallocate the table to a larger size. This concerns only v1.3.0 and v1.3.1 because these versions features the new robin-hood hashing I implemented which features the bug.
Anyway, I have integrated the fix in the new version I have been working on which I am gonna release asap. I'll let you know as soon as the release is there.
Guillaume
The branch bifrost-fileAsQuery
should have the fixes needed if you wanna try it now. I need to update the README, the Changelog and I still have to do a few more tests so the official release won't come before tomorrow.
README for that branch should be fairly up to date now.
v1.3.5 has been released, I am confident it solves your issue. I am closing this issue but feel free to reopen if you encounter additional problems.
Guillaume
Thanks Guillaume. I confirm that the current version works as intended. Volker
Great program. Thanks. I was testing a different program (https://github.com/tischulz1/plast) that makes use of bifrost and ended up with an odd error. Could you please take a look?
Here is my odd sequence:
Here is how I use it: mkdir fa cd fa wget 'ftp://hgdownload.cse.ucsc.edu/goldenPath/eboVir3/bigZips/160sequences.tar.gz' tar xzf .gz \rm .gz cd ..
ls fa | sed -e "s#^#fa/#" > slst
../bifrost-1.2.1/build/src/Bifrost build -o mygfa-1.2.1 -r slst -t 10 -n -v >& err-1.2.1 ../bifrost-1.3.0/build/src/Bifrost build -o mygfa-1.3.0 -r slst -t 10 -n -v >& err-1.3.0
cat slst | sed -e '$a gnm.fa' > slstF ../bifrost-1.2.1/build/src/Bifrost build -o mygfa-1.2.1F -r slstF -t 10 -n -v >& err-1.2.1F ../bifrost-1.3.0/build/src/Bifrost build -o mygfa-1.3.0F -r slstF -t 10 -n -v >& err-1.3.0F
Outcome: The plast example (ebola sequences) works fine with 1.2.1 and 1.3.0. Then I add my odd sequence to the fa directory. 1.2.1 is still happy. But 1.3.0 hangs on writing the index.
Multiple testing of similar cases on Fedora 39 Linux. I think this happens in IO.tcc, but you will know better. Thanks, Volker