amkwong
commented
4 years ago We managed to track down both of these errors, and resolved one.
The first error (points bunching together in a single column) was due to our code assuming that all gene names are unique (with unique TSS values), and grouping data points together based on this. The view that showed the error contained multiple genes (with unique ENSG gene IDs) which had the same gene name (RF00003 in our data set; also known as U1), which led to a thrown exception in our sorting code.
For now, we will sort this view based on TSS, which our code will assume to be unique. This part of the issue is assume to be resolved for now.
The second problem lies in gnomad, which accepts short gene names instead of ENSG numbers - this causes problems when the gene name maps to more than 1 ENSG number. This will need to be fixed in gnomad and LocusZoom.
Reported by @avsmith . Assigning to @amkwong for triage. Since the feature to query GNOMAD data is pretty new, feel free to bring me in for debugging if needed.
Description
Certain PheGET variant views are showing incorrectly.
Anomalies occur when switching between different TSS distance queries:
Steps to reproduce
Search for rs7212 and all the points bunch along the x-axis. Example link: https://eqtl.pheweb.org/variant/1_145992816/?minimum_tss_distance=-1000000&maximum_tss_distance=1000000
The view above starts at the largest TSS distance. Click on all the options, then return to the largest. You should see a gnomad error. According to the devtools console, the problem is a malformed query.