cBioPortal / cbioportal

cBioPortal for Cancer Genomics
https://cbioportal.org
GNU Affero General Public License v3.0
632 stars 484 forks source link

Black dots in sample circles on genie patient #7880

Closed alisman closed 4 years ago

alisman commented 4 years ago

https://genie-private.cbioportal.org/patient?studyId=nsclc_genie_bpc&caseId=GENIE-DFCI-008085#

image.png

adamabeshouse commented 4 years ago

@alisman so what's happening here is those dots represent a sample which doesnt seem to exist on the patient: GENIE-DFCI-008085-6295

because the sample doesnt exist, the code can't find a number to label the dot with, so defaults to -. If you mouse over you'll see there's some issue fetching data too.

adamabeshouse commented 4 years ago

@inodb any thoughts on this?

jjgao commented 4 years ago

Interesting, so it's data issue. In the error message could you add the sample ID? Once that's done, maybe we can close this? @alisman

inodb commented 4 years ago

I imagine the issue is that we do have a sample/sequencing collection date for that sample, but no sequencing data imported. Will share with Brooke and Ritika

inodb commented 4 years ago

@adamabeshouse it still might make sense to handle this a little more elegantly. E.g. one can imagine in the future the use case where sample acquisition has been done but there is not corresponding mutation data yet. Then the dot would show up in sample acquisition, but we wouldn't have mutation data imported yet. Maybe we can show a blue dot and when you mouse over it will show the regular timeline tooltip (instead of the sample tooltip)?

adamabeshouse commented 4 years ago

@inodb I agree we should handle it more elegantly. I feel that making it not look like a sample would be confusing, though. Here are a couple ideas: (1) Use a question mark instead of a number (but could be confusing) (2) Assign each unknown sample id a unique number (i.e. the last known sample's number + 1) in the SampleManager in a (temporarily) persistent way. Downside is that the numbers no longer would correspond at all to temporal order, but this maybe is a rare enough case that it doesn't really matter.

inodb commented 4 years ago

@adamabeshouse maybe (1), but continue to use -? Question mark feels more "alerting", whereas it doesn't necessarily have to indicate a problem. I guess using only one symbol wouldn't handle the case of having multiple samples with missing mutation data very well, but it's kind of a corner case anyway

adamabeshouse commented 4 years ago

@inodb gotcha, so we'll keep - but make the tooltip not show any error. But what should the tooltip say? @jjgao

inodb commented 4 years ago

It can be the same as the other tooltips on the timeline (don't show the sample specific tooltip, but rather the default one)