Describe the bug
The main platform, hosted at https://cellxgene.cziscience.com, recently broke in several ways. Please do tell me if this should be reported elsewhere. So far, I have noticed two key issues:
- Differential expression fails to return results.
- The platform fails to display genes present in data, and appears to be biased toward miscellaneous RNA species.
To Reproduce
This example is for a single dataset I happen to have on hand, but I see the same issues in spot checks in other datasets.
Steps to reproduce the DE bug:
- Go to the HYPOMAP dataset. Open the dataset using "Explore".
- Select any two distinct populations of cells.
- Press "see top differentially expressed genes."
- The outputs under "Pop 1 high" and "Pop 2 high" read
No genes to display.
Steps to reproduce the present genes bug:
- Download the HYPOMAP dataset. Consider the top genes by total expression, or a standard marker (SNAP25). These genes will not show up.
- So far as I can tell, this is not a split by
feature_type. Based on spot checks, some protein-coding genes (OR4F16, HES4, ISG15, etc.) appear on the platform.
- Based on spot checks, the following features do not appear on platform:
processed_pseudogene, snRNA, miRNA, unprocessed_pseudogene, misc_RNA (although there is a vast number of undifferentiated Y_RNA entries), transcribed_unprocessed_pseudogene (although SLC35E2A shows up), rRNA_pseudogene, rRNA, Mt_tRNA.
- The following features do appear:
lncRNA (but not all: a vast number of LINCs show up, but not not MALAT1), snoRNA (but not all: SNORA72 shows up but SNORD118 does not). I stopped looking here, as these categories cover most of the genes.
Expected behavior
- DE should always work.
- Genes present in the data matrix should always appear.
Screenshots
Today's version of the dataset (f91ae570a42434e8066d0b512584d711), and for that matter the October one I had on hand (not shown, dd5ffb49acfe2cb437d2fdee541e62df), have the correct genes (here, sorted by total expression).
These genes do not show up under the gene search, whether using gene name or ID.
Various other genes show up.
DE between astrocytes and oligodendrocytes in donor sample f5sVM produces no results. This also shows up with all other cell populations in all datasets I have tested.
Version (please complete the following information):
- Desktop or hosted?: Hosted
- Browser (if hosted): Safari, Chrome
- Version [e.g. 0.13.0]: ?
Additional context
Describe the bug
The main platform, hosted at https://cellxgene.cziscience.com, recently broke in several ways. Please do tell me if this should be reported elsewhere. So far, I have noticed two key issues:
To Reproduce
This example is for a single dataset I happen to have on hand, but I see the same issues in spot checks in other datasets.
Steps to reproduce the DE bug:
No genes to display.Steps to reproduce the present genes bug:
feature_type. Based on spot checks, some protein-coding genes (OR4F16, HES4, ISG15, etc.) appear on the platform.processed_pseudogene,snRNA,miRNA,unprocessed_pseudogene,misc_RNA(although there is a vast number of undifferentiated Y_RNA entries),transcribed_unprocessed_pseudogene(although SLC35E2A shows up),rRNA_pseudogene,rRNA,Mt_tRNA.lncRNA(but not all: a vast number of LINCs show up, but not not MALAT1),snoRNA(but not all: SNORA72 shows up but SNORD118 does not). I stopped looking here, as these categories cover most of the genes.Expected behavior
Screenshots
Today's version of the dataset (
f91ae570a42434e8066d0b512584d711), and for that matter the October one I had on hand (not shown,dd5ffb49acfe2cb437d2fdee541e62df), have the correct genes (here, sorted by total expression).These genes do not show up under the gene search, whether using gene name or ID.
Various other genes show up.
DE between astrocytes and oligodendrocytes in donor sample
f5sVMproduces no results. This also shows up with all other cell populations in all datasets I have tested.Version (please complete the following information):
Additional context