Transcription profiling by high throughput sequencing of five different organs of wheat (leaf, shoot, root, spike, and grain) at three developmental stages each
RNA-Seq mRNA baseline
Organism:
Triticum aestivum
Reference(s):
25853487
(Filter by genes in paper)
No change to top banner, tabs, or other tab content proposed
We get to this page after a search that matches one gene.
Proposal: after clicking on the heatmap rows in multiexperiment results, you go to this view too instead of the gene page
TODO: Should we do something if a search is for something that isn't a gene identifier, and matches one gene in this experiment?
Proposal: after clicking on the heatmap rows in multiexperiment results, you go to this view too instead of the gene page
TODO: Should we do something if a search is for something that isn't a gene identifier, and matches one gene in this experiment?
Genes
TRIAE_CS42_2BS_TGACv1_146768_AA0472460✖
Keep the current semantics - if you want to see gene specific results for a different gene, you have to add the new one and remove the previous one
(most specific does nothing if there's one gene)
Expression value
Data units
Experimental variables
Selected: 15 / 15
There might be very many assay groups and the UI still has to work well. Possibly: flip the charts so that different experimental variables are on x axis?
Switch to heatmap view Alternative proposal: just show the heatmap on top instead of switching to it
See results for this gene in other experiments Unnecessary but proposed - links to the gene page
See results for this gene in other experiments Unnecessary but proposed - links to the gene page
Gene annotations
+Attributes -> annotations
Include also Reactome pathways
Do not include "protein coding"
Maybe not on top - now it looks like you will filter results you already have
Similar: on the front page for an experiment have ready made categories that are enriched, facilitate searching for them
Unnecessary but proposed. When you click on these, the gene query box gets populated, and you see the gene you are looking at plus some context e.g. the rest of the pathway the gene is on. It's meant to reinforce that the user is looking at results for one gene, and might be useful - inspired by "filter by genes in paper". Cons: competes with the same information on the gene page.
serine-type endopeptidase inhibitor activity
integral component of membrane
Biotype: protein coding
Expression values
Switch this to the log scale!
Users are used to it, the boxplots getting squashed are fine, etc.
We already have this, when you "switch to boxplot view"
Per transcript expression
This is transcript data for the gene as the pipeline produced, quantile normalized and then aggregated into quartiles like we currently do for gene expression. The scatter points are results for individual assays.
The example got handpicked to be vaguely interesting - the dominant transcript switches across assay groups.
Other possible cases the UI should be able to present:
- one transcript (e.g. because a gene has only one exon)
- many transcripts, one is always dominant
- many transcripts, only one is ever dominant but not across all assay groups
Per transcript expression all on one chart
You can jitter the points slightly so they're not collinear - helps distinguish what's going on if there is a lot of assays per assay group
Proposal: also show the boxplots on this chart
This has the upside of being useful even the dominant transcripts are trivial or uninteresting, also the user can draw their own conclusion about what transcript is dominant.
Relative Isoform Usage
Another idea with less data. Aims to show how many transcripts there are, which transcript is dominant (most blue in a column) and whether this changes across assay groups.
Fake data, columns should add up to 100% vertically.
TODO: other visualisations? Maybe a "stacked percentage column"?