problems with OMA

Question

[jabc]

OMA finished the all agains -all phase but fails to produce output files and returns the following error:

***
Storing results
Error, 'array selector out of bounds'

Also several warnings are written regarding syntax error:

# Anopheles_gambiae_withid vs Ixodes_scapularis_withid done, (24193/24193)
syntax error:
[4921, 21077, 538.0622436, 73, 56..202, 48..192, 85.84397648] ]):

Answer 1

[Christophe Dessimoz]

It is difficult to tell for sure based on your output alone, but it seems that at least one species pair file might be corrupt, probably because you had more than one process simultaneously writing out to the same file (not sure why this happened - if you run parallel processes as described in the OMA manual, such clashes should be avoided).

The affected pair is the one that immediately follows "Anopheles_gambiae_withid vs Ixodes_scapularis_withid" (I cannot tell the full list of genomes from your output). If you are unsure, zgrep all files in Cache/AllAll for the offending string and delete the corrupted file.

[jabc]

Thanks for the answer, I found and deleted the corrupted files and rerun OMA. This time there are no syntax error warnings: but still no output files are produced with the sole exception of Map-SeqNum-ID.txt. Any ideas?

Pediculus_humanus_withid vs Pediculus_humanus_withid done, (7231/7231)
# Pediculus_humanus_withid vs Tribolium_castaneum_withid done, (28558/28558)
# Tribolium_castaneum_withid vs Tribolium_castaneum_withid done, (44280/44280)

****

Storing results

Error, 'array selector out of bounds'

(END) 

[Christophe Dessimoz]

Unfortunately, it's likely that some other files are corrupt (the syntax might happen to be "correct", which won't throw an error, but the content will be erroneous). This can happen if you have multiple processes trying to write to the same file at the same time. If you want to run multiple processes, use the parallelisation options as described in the manual (either using the -n option if you run OMA on a single multicore machine or using job arrays with supported schedulers).

I'm afraid the best course of action is to restart the all-against-all from a clean state. If you exported genomes including all-against-all computations from the OMA browser, I suggest that you redownload a clean archive. If you are only using your own data, delete the "Cache" directory and restart using the proper parallelisation options.