Empty BAM files generates an error in the gridsearch_filter rule about swgs-absolutecn

[Phil9S]

Thanks Thomas. I'm curious as to what circumstances would lead to a situation where you might attempt to process an empty BAM file? Did you purposefully invoke this error or did it occur by chance? Either way it's best to resolve it and attempt to make the pipeline as fool proof as possible. I'll update this issue once I have implemented a fix.

[TBradley27]

Hi Phil,

I will have to confess and say it was laziness/negligence on my part. I had a batch of samples to process. I ran the necessary alignments, but because I was working late, I forgot to check which files were empty/non-empty, and simply plugged all the bam files into the pipeline.

I guess the onus could be put on the user to do this themselves.

Thanks

[Phil9S]

Resolved in release v1.1.0

[TBradley27]

Hello,

I am afraid that this is still an issue for bam files which are non-empty and valid but nonetheless produce copy number plots with very high variance and NA values for the loess.span and loess.family columns

e.g.

[Phil9S]

Hi Thomas, does this prevent the pipeline from running and invoke an error or does it just generate a poor fit due to low reads?

[TBradley27]

Hi Phil, sorry for the delayed reply - this prevents the pipeline from finishing completely.

The source of the problem is that the NA values causes the R to treat some of the columns as logical vectors rather than numeric which causes an error when R tries to combine the RDS files