Add genetic demultiplexing component

[wxicu]

Issues to be solved by adding genetic demultiplexing component:

1. Can we use public available containers, e.g. these provided by biocontainers, or we need to define these by our own?
2. How can I add data for the test scripts? I have used the files demuxafy_test_data, but sill need to upload more files.

Thanks!

[wxicu]

Hi! Thank you for the comments and hope you had a nice Christmas!

• resources_test_scripts are added
• For cellsnp and bcftools I have changed the docker configuration to ubuntu:latest, but it can be changed easily if we agree on specifying a specific ubuntu version.
• For freemuxlet and demuxlet: It doesn't work on my Macos arm64 architecture currently to either build container from dockerfile or by openpipeline (it used to work...). I also tried Dries configuration. All return segmentation fault. I am still looking for a solution. But it works on other platforms...
• The source code is independent of demuxafy. I only use some test datasets from it.
• For the input: The genetic demultiplexing tools have the bam input in common except for vireo. There are some differences in other input requirements. I will try to make a table to discuss in the meeting later.
• For the output: Although different tools have different output, but they all have a file (tsv, csv, etc) to save the genetic demultiplexing result(singlet/doublet/unassigned, and donor identity). In the nextflow pipeline I implemented before, I have a process to summarize all the genetic demultiplexing results into a single table. Is is better to have such a summarizing step at the end after running different tools or to convert to a specific format every time we run a component? But yes again, we can discuss the details in the meeting.

[wxicu]

Hey hey, a small update:

• For the output format, each component has an output file in the same format called assignment.tsv
• For the input format: they have different input so it is difficult to let them have the input in the same format

[rcannood]

Hey @wxicu! Thanks for the update :)

• I ran the resources test script. Unfortunately, the resulting files are quite big:

  

  • Is it possible to subset the bam file?
  • In addition, would you be able to use the fa/fa.fai files that are already available at resources_test/cellranger_tiny_fastq/cellranger_tiny_ref/fasta? These are much smaller since they are subset to contain only chr 21, and it's already on S3 so it wouldn't consume any additional storage space.

• demuxlet and freemuxlet are both missing a file called summary.R. Where is this R script located?

[rcannood]

Also @wxicu, we've gone back and forth a few times now. Would you like to have a one on one meeting so we can fix the remaining issues and finalize the PR asap? :)

[wxicu]

@rcannood Thanks for viewing the PR again!

• Actually, I think the bam file should be already on the S3, as it is also used by https://github.com/openpipelines-bio/openpipeline/blob/feature/count_demultiplexing/src/deconvolution/demuxlet/test.sh here.
• Souporcell didnt work with this genome reference because vatrix could not find sequence in FASTA. So I just subsetted the bam and genome by chr1 and uploaded to the dropbox. The download script has been changed correspondingly. Hope the data is smaller now!
• Oh Sorry I forgot to upload it! Just Done.

[rcannood]

Not sure what's going on with this merge. Closing and reopening.