Minia exits with error in "Debloom: cascading" stage

Question

[Shellfishgene]

Hi!

I'm trying to run minia for an assembly, but I get an error in the "Debloom: cascading" step:

[DSK: Collecting stats on read sample   ]  100  %   elapsed:   0 min 21 sec    estimated remaining:   0 min 0  sec   cpu:  147.4 %   mem: [ 51,  51,  51] MB
[DSK: Pass 1/1, Step 2: counting kmers  ]  100  %   elapsed:  12 min 28 sec    estimated remaining:   0 min 0  sec   cpu:  162.7 %   mem: [264, 5554, 5917] MB
[MPHF: initialization                   ]  100  %   elapsed:   3 min 24 sec    estimated remaining:   0 min 0  sec   cpu:   36.2 %   mem: [2409, 2409, 5917] MB
[MPHF: build hash function              ]  100  %   elapsed:   4 min 52 sec    estimated remaining:   0 min 0  sec   cpu:   82.8 %   mem: [2825, 2825, 5917] MB
[MPHF: assign values                    ]  99   %   elapsed:   0 min 22 sec    estimated remaining:   0 min 0  sec   cpu:   99.8 %   mem: [1983, 1983, 5917] MB
[MPHF: populate                         ]  100  %   elapsed:   1 min 19 sec    estimated remaining:   0 min 0  sec   cpu:   72.3 %   mem: [424, 424, 5917] MB
[Bloom: read solid kmers                ]  100  %   elapsed:   0 min 13 sec    estimated remaining:   0 min 0  sec   cpu:  252.1 %   mem: [534, 534, 5917] MB
[Debloom: build extension               ]  100  %   elapsed:   1 min 8  sec    estimated remaining:   0 min 0  sec   cpu:  229.4 %   mem: [544, 544, 5917] MB
[Debloom: finalization                  ]  100  %   elapsed:   0 min 4  sec    estimated remaining:   0 min 0  sec   cpu:  213.9 %   mem: [448, 449, 5917] MB
[Debloom: cascading                     ]  0    %   elapsed:   0 min 0  sec    estimated remaining:   0 min 0  sec   cpu:   -1.0 %   mem: [448, 448, 5917] MB 
EXCEPTION: cannot open cfp No such file or directory

What does that mean? Not sure how to debug this. This is the command I ran:

minia -in B0928.files.txt -kmer-size 31 -max-memory 16000 -out-dir B0928

The text file points to two fastq.gz files. This is Minia 2.0.2.

[Rayan Chikhi]

Hi, thanks for reporting this bug. I'd like to check one thing with you first:

• Can you reproduce this problem when you run an assembly without the -out-dir parameters, i.e. by running minia inside the B0928 working dir?

[Shellfishgene]

Hi Rayan,

actually I noticed of 10 jobs only about half had the problem. I started them again with the exact same settings, and most of them finished. The one's that did not I just restarted a third time and now all worked. The problem seems to be intermittent, maybe it's caused by our cluster rather than Minia itself.

Answer 1

[rchikhi]

Actually, a colleague (@edrezen) told me of the existence of a bug that's likely causing what you experience. A file "cfp" is created in the working directory during Minia's execution, so if you have multiple runs in the same working directory, it's likely that one of the runs will overwrite another's cfp. He told me that this bug will be fixed soon :) I'll let you know. For now, as a temporary fix, please avoid running multiple Minia's at the same time in the same working directory.

[Shellfishgene]

Ok, that makes sense, so when I was running fewer Minia jobs the chance that there was that file conflict was lower, that's why it worked. Thanks!