Closed bernardo1963 closed 5 months ago
I was able to run this locally with v2.1 without any consensus errors/warnings. I suspect this is the same as issue #247 and is fixed by this commit: https://github.com/marbl/canu/commit/c61ebbb7a5f90abb9c034650e2fd95642138de31. In that case, you'd need to build canu locally and replace the utgcns binary in your conda installation of verkko. Before you do that though, I'd like to verify if that's the case, can you share the cnspack for one of the failing partitions so I can run it locally?
Dear Serge, Thank you for your help. I uploaded to the same GoogleDrive folder all part005 files (including part005.cnspack ). part005 is the one that produced the fatal error.
Thanks, I confirmed the crash and that the reference commit above fixes the issue. It will be in the next verkko release but in the meantime, if you build canu from tip and replace the utgcns binary, it should address your crash.
I saw a message about building canu but din't see the post here, I assume you resolved the issue? It looked like an older version of git that couldn't initialize the submodules (https://github.com/marbl/canu/issues/1747). It should be possible to run git submodule update --recursive
to get the modules up to date to build canu.
Dear Sergey, Thank you for your help and care. I eventually succeeded, and got my verkko assembly. My server has CentOS7, which has an old gcc (4.8.5) and an old git (1.8.3.1); I think both cause errors while compiling canu. As others may stumble with the similar problems, I copied what I did below:
mkdir /home6/tools/temp_git2
cd /home6/tools/temp_git2
scl enable devtoolset-9 bash
gcc --version
#gcc (GCC) 9.3.1 20200408 (Red Hat 9.3.1-2)
scl enable rh-git227 bash
#git version 2.27.0
git clone https://github.com/marbl/canu.git
#remove the original unitigConsensus.C
rm /home6/tools/temp_git2/canu/src/utgcns/unitigConsensus.C
#now move the new commit of unitigConsensus.C to the same folder
cd /home6/tools/temp_git2/canu/src
make -j 32
#find the executable utgcns inside canu
cd /home6/tools/temp_git2/canu
find . -type f -name "utgcns" -exec ls -lh {} \;
#it is here: /home6/tools/temp_git2/canu/build/bin/utgcns
#Here are the binaries I should replace inside verkko
mv /home6/tools/miniconda3/envs/verkko/lib/verkko/bin/utgcns /home6/tools/miniconda3/envs/verkko/lib/verkko/bin/utgcns.old
mv /home6/tools/miniconda3/pkgs/verkko-2.1-h45dadce_0/lib/verkko/bin/utgcns /home6/tools/miniconda3/pkgs/verkko-2.1-h45dadce_0/lib/verkko/bin/utgcns.old
#copying the new utgcns (commit) to verkko
cp -p /home6/tools/temp_git2/canu/build/bin/utgcns /home6/tools/miniconda3/envs/verkko/lib/verkko/bin/
cp -p /home6/tools/temp_git2/canu/build/bin/utgcns /home6/tools/miniconda3/pkgs/verkko-2.1-h45dadce_0/lib/verkko/bin/
#now verkko run w/o errors
Thanks for the update.
Hi, I am using verkko 2.1 (bioconda verkko bioconda 2.1) in a local machine with 208 threads and 2 TB RAM, running CentOS7. I am assembling the Y chromosome of HG002 using subsets of the HiFi and ONT UL reads, and I am repeatedly getting a segmentation fault error during the consensus step. The HiFi and ONT reads were obtained as follows. I used minimap2 and seqtk to get the Y-linked reads from HiFi and ONT reads, and assembled them with verkko. Several variations worked well, but one paticular read dataset (HiFi + ONT) is always causing a fatal error (I uploaded these reads to https://drive.google.com/drive/folders/1kp4zPwQ27rNNzvXeM-3MdUT5slNXiLUH?usp=sharing). 2024-06-20T113029.669398.snakemake.log The command line I am using is:
I attached the snakemake log file. The error message follows:
Digging a bit more, I found no error message in ./7-consensus/buildPackages.err , but there is a detailed report in 7-consensus/packages/part005.err , as follows in the end of this message. I obtained the reads twice (using seqtk)), with the same error result. So now I am stuck. Any suggestions?
thanks, Bernardo