Add ability for rust uniwig to create output files from input .bam files

[donaldcampbelljr]

Some work accomplished with PR #40.

We would like to use this code as a drop in replacement for bamSitesToWig.py from PEPATAC: https://github.com/databio/pepatac/blob/master/tools/bamSitesToWig.py

bamSitesToWig.py creates three files as output:

1. • [ ] exact.bw
2. • [ ] smoothed.bw
3. • [ ] shift.bed

Currently, uniwig can take an input file of either: bed, narrowPeak, bam

and create an output of: wig npy bedGraph bw (via an intermediate bedGraph conversion)

Some items to accomplish for this task:

• [ ] add bed output support if given a non-bed-like file as input (i.e. bam).
• [ ] Original bamSitesToWig adjusts positions based on flags. Noodles does support Flags but on first glance they are named differently and must be reviewed.
• [ ] I believe PEPATAC pipeline only needs starts counts? So add ability to only output any counts the user wishes instead of always outputting all three (starts,ends, core).
• [ ] Scaling is used in bamSitesToWig.py. Currently, uniwig does not scale counts. Should it?
• [ ] Smooth vs Exact, currently uniwig only offers the smooth option, though the user can set smoothing =0. Is this the same output as the exact code? If so, why were two perl scripts used? For original counting algorithms, see: https://github.com/databio/pepatac/blob/master/tools/cutsToWig.pl https://github.com/databio/pepatac/blob/master/tools/smoothWig.pl

Nice to have:

• [ ] if associated .bai does not exist, have uniwig create it (using noodles crate)

[donaldcampbelljr]

Working proof of concept in #40 , however, it uses an intermediate BedGraph file written to disk.

Therefore, we are exploring an alternative method in #47 which streams values directly to bigtools bw writer.

However, some challenges remain, namely:

• [ ] when spawning separate threads to handle consumer/producer workflows, I see issues with error handling where an error does not seem to close the producer, causing the consumer to "hang"