Failing to parse "GNOMADE_POPMAX"

[Jakob37]

Failing to parse "GNOMADE_POPMAX"

[northwestwitch]

The issue seems to be that that key simply say which of the sub-populations had the highest frequency. And then gnomADg_AF_popmax specify the frequency for that sub group.

I see, but then let's fix it so it grabs the value only if it's a number! I'll open a PR!

[northwestwitch]

nfe as non-finnish europeans? And why does it take the place of a number? I'm going to look to their repo 🤔

[Jakob37]

nfe as non-finnish europeans? And why does it take the place of a number? I'm going to look to their repo 🤔

I think "nfe" refers to a subgroup within gnomad (afr is another I have seen - african). You are probably right - non-finnish europeans. The issue seems to be that that key simply say which of the sub-populations had the highest frequency. And then gnomADg_AF_popmax specify the frequency for that sub group.

[northwestwitch]

I was wondering which values gave error for instance in the demo app, so I've added a warning message for debugging and I saw this:

Perhaps we should try to collect numbers present within parenthesis

[Jakob37]

Perhaps we should try to collect numbers present within parenthesis

Hmm. I guess it depends on where that deleterious(0) value is produced. Often better to error than ending up with unexpected and unwanted values, if not part of the "standard format". Do you know how your Gnomad values are inserted into the VCF? I don't think we have these values (i.e. deleterious(0)) in our data.

In our case we (long before I started working at CMD) have prepared the Gnomad data in an indexed file. This is inserted directly on matching variants using VEP's -custom flag.

1       10108   rs62651026      CAACCCT C       46514.3 RF      AF=0.00100908;nhomalt=0;popmax=afr;AF_popmax=0.0035461
1       10109   rs376007522     AACCCT  A       89837.3 RF      AF=0.0641892;nhomalt=0;popmax=amr;AF_popmax=0.1

[Jakob37]

I think the keys "popmax" and "AF_popmax" are selected by us here. Would be nice to be able to continue using them for a bit longer though.

Soon we will transition to Gnomad 4 where they call things "grpmax" though instead of "popmax". That is probably something you will see / already have seen as well.

[northwestwitch]

Do you know how your Gnomad values are inserted into the VCF? I don't think we have these values (i.e. deleterious(0)) in our data.

I have no idea, it's very old data and perhaps current cases don't have that format any more. You see those error warnings only with a recent scout version because of this relatively new change in transcript parsing. I'm still not sure it's a bug, but all those error message are annoying for sure..

[northwestwitch]

Soon we will transition to Gnomad 4 where they call things "grpmax" though instead of "popmax". That is probably something you will see / already have seen as well.

No problem, we can add the keys!

[Jakob37]

I have no idea, it's very old data and perhaps current cases don't have that format any more. You see those error warnings only with a recent scout version because of this relatively new #4456 in transcript parsing. I'm still not sure it's a bug, but all those error message are annoying for sure..

OK, I'll have to look a bit more that the code to understand what is going on I think.

It seems to me that Scout is picking the wrong value to try getting the max population frequency, which means we end up with no value there. If so, just skipping them would not fix the issue.

I'll check a bit closer. Will be back ...

[Jakob37]

Aha, or I see, it is perhaps just trying to add the group name in addition to the frequencies. Then the change makes sense. I'll test the PR!