Closed gwaybio closed 2 years ago
Logs for that site look normal, so I don't know. Working at genome scale I wouldn't be surprised to lose a site or two to a corrupted/incomplete file. My suggestion would be to update the recipe so it skips ParserErrors here, but log and check how many times it happens? Or to avoid having something you have to check, what about if there is a ParserError it initiates a counter and if the counter hits 5 (a number I arbitrarily chose right now) then it stops the run.
Sounds good, thanks @ErinWeisbart - I added broadinstitute/pooled-cell-painting-profiling-recipe#79 to track this fix.
I'm now running CP186 through the recipe and it proceeded past CP186D-Well2-16 no problem but looks like it errored on CP186G-Well2-66 (Unfortunately, I don't have the stack trace).
And now on another run also passed CP186D-Well2-16 just fine but errored at CP186D-Well5-31 (which it passed fine last time). So another suggestion that it's stochastic.
Thought I had fixed the error handling (spoiler alert: I hadn't) and ran again and it errored at the same site this time and I think at least this time it has to do with the files not being unarchived. (Though FWIW I believe the file restoration error is stochastic..)
I believe switching to naked excepts from https://github.com/broadinstitute/pooled-cell-painting-profiling-recipe/commit/cbde4011ace3476005b87d6661a9996a0129aa61 handles this error (though I also re-unarchived the files before running). Regardless, I'm closing this as this is no longer a problem for CP186 as we've progressed beyond this point in running the recipe.
A new error appeared!
@ErinWeisbart - any idea what's going on? It looks like the foci data for site
CP186D-Well2-16
is corrupted somehow. Would you know off the top of your head why?