Catch MACS error about chipseq

[orzechoj]

A quick and ugly fix is to add errorStrategy = 'ignore' to the macsjob in base.config. But then we don't see which macs jobs have failed (and we ignore other errors in the macs jobs too).

[apeltzer]

Is there a special error code that MACS generates in such a case? That would be interesting to catch and issue a warning for that sample.

[ewels]

You mean an exit code? Not sure. I was saying to @orzechoj that we could implement a similar strategy to the one that we have in the rnaseq pipeline though, where we check the log files:

https://github.com/nf-core/rnaseq/blob/d64b7f536b09b400c8d46533b150c4f05843b556/main.nf#L591-L595

https://github.com/nf-core/rnaseq/blob/d64b7f536b09b400c8d46533b150c4f05843b556/main.nf#L530-L549

[apeltzer]

Yeah, exit code was in my mind actually. I thought the failing MACS process issues an exit code != 0, therefore causing the pipeline to fail. If it just pushes something on stderr, we can implement it the way you proposed - nice 👍

[ewels]

The problem with the exit code is that it probably doesn’t differentiate for why it fails - we only want to catch these specific failures. I could be wrong though!

[apeltzer]

I fear so, too! If I find some time I'll have a look.

[orzechoj]

The exit code is 1, which isn't so informative. But stderr has this error message

Too few paired peaks (0) so I can not build the model! Broader your MFOLD range parameter
may  erase this error. If it still can't build the model, we suggest to use --nomodel and 
--extsize 147 or other fixed number instead.

[drpatelh]

Closing in favour of #128