Isoseq cluster2 failed when running with more than 30M consensus reads.

[7kylin]

hi，all： My problems are as follows: Operating system CentOS Linux release 7.4.1708 (Core)

Package name isoseq 4.0.0 (commit v4.0.0)

Using: pbbam : 2.4.99 (commit v2.4.0-16-g5cc6e4b) pbcopper : 2.3.99 (commit v2.3.0-14-g5ac5693) pbmm2 : 1.11.99 (commit v1.11.0-1-g1b5a417) minimap2 : 2.15 parasail : 2.1.3 boost : 1.77 htslib : 1.17 zlib : 1.2.13

Describe the bug terminate called without an active exception /opt/gridengine/default/spool/tjcompute133/job_scripts/336846: 行 2: 34707 (core dumped) isoseq cluster2 flnc100.bam transcripts.bam

To Reproduce I qsub this step with vf=500G,p=50 ,but it still exited with this error.

Expected behavior How can I run cluster with 30M consensus reads?

Thanks for your reply.

[armintoepfer]

We can't help you given the information you've provided. Have you tried clustering fewer reads? When does it start breaking? We've run it with up to 500M HiFi reads.

[7kylin]

We can't help you given the information you've provided. Have you tried clustering fewer reads? When does it start breaking? We've run it with up to 500M HiFi reads.

Yes, The program running normally with less than 20M reads. Can you tell me how much memory and cpu you use? I will try it later. Or what is needed for this program is GPU? Thanks a lot.

[armintoepfer]

We run with -j 128 for those benchmarks

#HiFi	Wall time	CPU time	Peak RSS
500M	1d 15h	38d 20h	954 GB
250M	11h 14m	28d 5h	317 GB
125M	6h 23m	10d 4h	127 GB
63M	2h 5m	4d 11h	69 GB
31M	58m 44s	2d 1h	50 GB

[7kylin]

We run with -j 128 for those benchmarks

HiFi Wall time CPU time Peak RSS

500M 1d 15h 38d 20h 954 GB 250M 11h 14m 28d 5h 317 GB 125M 6h 23m 10d 4h 127 GB 63M 2h 5m 4d 11h 69 GB 31M 58m 44s 2d 1h 50 GB

I have got it and thank you very much.

[armintoepfer]

What was the resolution?