Closed andrewhercules closed 5 years ago
Hi Andrew, yes the bucket-column mapping there is incorrect. This is fixed for the new release (i.e. v2 of prioritisation view + batch search) but I can cherry-pick the commits and push that with next mini-release:
Columns should be mapped (in order L to R, dropping 'unknown'/'rest of genome') as bucket: 1, 2, 3, 4, 7, 5, 6, 8
Also for Antibody, bucket 5 should be GO cell component - high confidence
.
Pushed mini-release with the fix.
I've come across another example of inconsistency between UI and API response for small molecules:
API response:
UI rendering:
(target: APP)
It turned out this is not a bug. Thanks @andrewhercules for the explanation. The correspondence between buckets and cells highlighted in purple in the UI is a weird one and confusing when compared to the API response.
The order of the cells in the table is bucket 1 bucket 2 bucket 3 bucket 4 bucket 7 bucket 5 bucket 6 bucket 8
We will convey this into our docs/Chris' blog post as users may assume the order of the cells in the UI is bucket 1, 2, 3, 4, 5, 6, 7, 8.
It may be worth showing in the UI which cell corresponds to which bucket.
Also, buckets 9 and 10 from the original paper have been either dropped or are not shown.
Observed Behaviour
On the target profile page for VCP, the tractability heat map shows that the target hits the following buckets:
In the image, it looks like the target hits buckets 5, 6, and 8. However, according to the .TSV file from Chris and the API response, it should hit buckets 4, 5, 7, 8.
Expected Behaviour
The heat map should correctly render the data from the API response and in this case, highlight the
PDB targets with ligands
,Active compounds in ChEMBL
,DrugEBIlity score 0 to 0.7
, andDruggable Genome
cells.