germline-filter run instructions about cerebra HOT 5 CLOSED

Hi Roshan thank you for your interest in the tool. Yeah you're absolutely right: the germline-filter module could be configured in a more tolerant manner. I'll add this to the list of things to do.

In the meantime, I think all --metadata needs to be is a comma separated values file with cell_id and patient_id headers. It could be as simple as this:

We plan on adding adnl documentation in the future but the basic idea is if you have a set of samples and a set of patients those samples came from, you can filter out the common variants found across all of the samples from a given patient. So for the above example, if cell1 and cell2 share 10 variants, those variants would be removed by germline-filter as they are common across all samples of patient1.

from cerebra.

Hi Lincoln,

Many thanks for the clarification regarding the metadata option.

I have a few further questions I hope you can clarify about the tool:

1. Should the cells VCF file be a joint called file of all cells or each cell contained in a seperate VCF with multiple files then passed in via --cells ?

2. Is the filter for shared somatic variants between the cells able to be adjusted or disabled?

3. On the Readme page it mentions the framework was designed for scRNA-seq, are there any specific filters the tool applies to adjust for the inherent technical noise of scRNA sequencing?

Best wishes,
Roshan

from cerebra.

Hi Roshan

1. Its probably best to keep vcf files separate. For example, if you have a directory full of vcf files named vcf_dir then for the path to germline vcf files directory: argument you would specify vcf_dir

2. If you dont want to use the filter you can just run the other two modules (count-mutations and find-aa-mutations) on an unfiltered set of vcf files. In other words, the two subsequent modules do not require vcf files to be germline filtered

3. As of right now we're not doing any filtering for potential technical noise. Its a very interesting problem but is perhaps a bit outside the scope of this package. Right now cerebra just summarizes variants found in a set of vcf files -- it doesnt make any judgments as to whether or not those variants are "real".

from cerebra.

hopefully this is addressed by b020ac8

from cerebra.

this should also be addressed by #74

from cerebra.