search

sigven / pcgr

Personal Cancer Genome Reporter (PCGR)
https://sigven.github.io/pcgr
MIT License
252 stars 48 forks source link

Request: Cohort analysis #93

Open adbeggs opened 5 years ago

adbeggs commented 5 years ago

Thanks for the great tool - very useful! Do you think there would be any chance of a summary HTML across a cohort of samples that have been through PCGR?

sigven commented 5 years ago

Dear @adbeggs, Thanks for getting in touch! And for bringing forward a very relevant issue, and that has been brought to my attention in several projects where I have utilized PCGR for sample annotation. A lot of people have been requesting similar functionality, but I have not yet been been able to prioritize it.

I think having a simple, basic report for a sample cohort would not require too much work. That said, I would ideally like some input in terms of what you consider most valuable in such a report. Importantly, when present with multiple samples, it brings forward a whole new set of opportunities for analysis (consider e.g. what is offered by maftools, which by nature is looking at multiple samples).

best, Sigve

adbeggs commented 5 years ago

I think similar to maftools - which I have used as well. Your report is very close to the one that the 100,000 Genomes project produces in the UK and that is excellent - I was wondering about a tiered germline module as well, as this would be useful as well? Also the ability to input CNA as VCF files rather than segmentation as lots of callers are moving towards this (I am thinking Canvas in particular)

sigven commented 5 years ago

Thanks for the feedback. Will definitely look into having CNA fed as VCF, I certainly agree this would be handy for several people.

For tiered germline variants, I have developed a stand-alone workflow (building upon PCGR) in a separate repository: Cancer Predisposition Sequencing Reporter (CPSR). Among other things, CPSR utilizes >35 different (cancer) gene panels curated through the Genomics England PanelApp. Example report here.

best, Sigve

sigven commented 4 years ago

Hi Andrew, Getting back to this. Do you have an example VCF with copy number calls? Sigve

adbeggs commented 4 years ago

testcnv.zip

Sure - try this - it is from the Illumina Canvas caller. I've been using your pipeline for Nanopore Promethion WGS data too and it works well!

adbeggs commented 4 years ago

Also, SV integration would also be useful - we've used Manta for Illumina calls and Sniffles for Nanopore... happy to supply data from these as well. We've tried to write our own pipeline but when we found yours it seemed rather like reinventing the wheel!

sigven commented 4 years ago

Hi,

Just released 0.9.0, and also included a FAQ section, where I make a few comments regarding multi-sample analysis, and that structural variants should be supported in the future. Will try to work on the VCF-based copy number calling (if time permits).

best, Sigve

dudious commented 3 years ago

Hi there,

Thanks for making this amazing tool, we are often doing multiregional sequencing where a cohort is actually just one patient, Being able to merge (without losing the extra dimension) the data table would be of great use in this scenario. An UPSET plot would be a nice visualization of this.

Best, Wouter

sigven commented 3 years ago

Hi Wouter, Thanks for reaching out. This could be a really useful extension (and strategic way further) for PCGR, to support multiple tumors for a given case, I guess either temporally or spatially (as in your case). I think this is a scenario that will become more frequent in the future. Let me know if you have detailed thoughts on other aspects of the output here. I think I also would need some more hands to move such a project forward, might be suitable for e.g. a Master's project.

kind regards, Sigve