Scout report improvements for Cancer view

Besides being able to know which filter was used to generate the report, I’d like to also be able to see a table of variants in the report. For rare diseases this doesn’t matter much, right? But for cancer view, it’d be great so the report contains the following information:

• Discard inheritance model reporting

• Discard GQ (?)

• Add AF of reported variant

• Squash variant table into a more compact form. e.g. we can remove allele depth and add AD to reference and alternative

• It still says “Affected”: Yes/No even for cancer variants. I’d like to have it as Tumor or Normal. Or just use “Phenotype”, Daniel: possibly with another name

• Discard Single sample family. No pedigree available. for cancer samples. There is no pedigree.

• Discard 1000G, always show gnomad? pop_freq_max? Daniel: Population frequencies are shown as annotated (and parsed on import). We can certainly ditch or deprecate 1000G from parsing, but if it is not annotated it will also not be shown. This is a general issue for Scout really, that could so slowly perhaps be resolved by only showing GnomAD assuming that essentially all cases should have it.

• Add bait set name to the report instead of “PANEL”. HFA: Could this be a value in Scout config? Also ultimately the information about panel according to https://clincial-genomics-hybrid-capture.readthedocs.io/en/latest/ in a nice format as BALSAMIC report. Daniel: this has been extensively discussed in other issues, and conforming to Scout panels will do much good for the overall experience. But if you want not only the gene panel for use with Scout display filtering, but also a capture panel linked, add it as a key to the load config and we’ll link to it from both the case page and show it on the case report. Compare e.g. the rank model links that we have now.

• Discard ancestry if empty (i.e. N/A)

• Discard parenthood if empty (i.e. N/A)

What do you think? @vwirta any more items I missed?

[dnil]

That’s a bit pessimistic, but you’ve got a point.

Cheers - you are very reasonable! I wouldn’t say pessimistic; just matter of fact on that part. No reason to introduce loads of conflicting code to fill that niche?

Fair enough, it is very hard to design a one for all solution. This seems like a tricky sitatution.

Until someone shows me otherwise, I’m going to remain convinced that is what your end users will want - at least in a few years when they have gotten some large scale genetics experience. Granted, depending on who they are and what they do, the may also want biomarkers, but again, that is very easily solved with a bypass mechanism. No manual triage needed for those!

[vwirta]

Some further ideas to add to report

• bait set used for enrichment (sequencing field says now ‘panel’, but would be more informative if we add the actual bait set)

To remove?

• In Pinned varriants: Do we still need to have the 1000G freqs? Would it be more informative with gnomAD(max)?
• In Other Manual Ranked Variants: would look better if freqs here are the same as for pinned variants