search

alekseyzimin / masurca

GNU General Public License v3.0
242 stars 35 forks source link

MaSuRCA-3.2.7 Restarting during CABOG stage #54

Open TimothyStephens opened 6 years ago

TimothyStephens commented 6 years ago

Hi,

I noticed that when you restart during the CABOG assembly stage the CA.mr.* directory is removed and recreated (this appears to be happening in the mega_reads_assemble_cluster.sh script). After looking through the runCA code it appears that this script can recover from a restart however the removal of the CA.mr directory means that after a restart all steps have to be recomputed.

Is this an intended behaviour? Am I incorrect in saying that the runCA script can restart after a crash? I ask because this stage (particularly the 1-overlapper step) can take a very very long time and I would prefer to not loose all progress on every restart.

Thanks,

alekseyzimin commented 6 years ago

Hi, restarting in CABOG stage works, make sure you re-generate assemble.sh before restarting. mega_read_assemble_cluster only deletes CA.mr... if the original frg files were deleted. There is one bug where if you kill CA run in overlap filter stage, (ovlStoreBackup folder created), it would not find overlapstore on restart and try to re-do overlaps -- in this case move the contents of ovlStoreBackup back to CA.mr..... This is fixed in upcoming 3.2.8 release, beta available on github.

TimothyStephens commented 6 years ago

Thanks. I must have forgotten to regenerate the assemble.sh.