broadinstitute / dropviz Goto Github PK
View Code? Open in Web Editor NEWShiny app for visualization, exploration of mouse brain single cell gene expression
Shiny app for visualization, exploration of mouse brain single cell gene expression
When new bookmarks are saved, then restoration works as expected.
For old bookmarks (?), the parameters are loaded and properly rendered, but the query panel remains orange. Pressing Update or resetting creates strange results.
Some gene symbols are not in the descriptions file (Drd1) even though the symbol is in the DGE.
Example Drd1 displays with auto-completion, but the server complains that the gene is unknown.
Cerebellum Cluster 7 currently is missing from the DropViz app data.
This is because the cluster_sheet was accidentally not included. Arpy created a replacement at /Volumes/broad_mccarroll/dropseq/data/clusters/atlas_ica/arpy_test/F_GRCm38.81.P60Cerebellum_ALT/cluster_sheets/Cluster_7.csv, but rerunning prep will cause bookmarking to break.
See email from Arpy on Apr 9, 2018 "discovered a missing "cluster_sheet" in the Cerebellum dataset, wanted to let you know".
Sometimes, particularly I think with a new or slow instance, the gene symbol restoration does not occur. But it is not reproducible. filter.vals$user.genes gets set correctly during onRestore, but input$user.genes does not get set during onRestored. Almost certainly this is a timing issue where valid values for user.genes have not been updated before onRestored is called. It may be related to the amount of work for display.
Is there a workaround? There's no way to know whether input$user.genes is blank because the user removed the values or because the values were not restored. This is almost certain related to starting the choices as NULL and then updating to use server side choices. Is there someway to prime the choices with a small set of values that will match?
See email April 9 from [email protected] with new CSS.
Also needs <meta name="viewport" content="width=device-width, initial-scale=1">
The code to download the R plot function fails on the server. The lines for the function are different and the code is reformatted so that "else" statements are dangling, resulting in a parse error. When I print a function interactively on the server, it works as expected and yet when shiny is running on the server I get a different result.
I think the solution is to use a different method. Instead of printing the function object, separate the function into a separate file, use a source('tsne-plot.R', local=TRUE)
and then just send that separate file with the data each time.
Lines 45 to 69 in b29f706
"Ento Peduncular" should be "Entopeduncular"
The three classes of "Ependyma*" should all be the same.
This should be fixed when #18 is dealt with.
Update the exp_sets.txt for Ento Peduncular.
Have Apry make changes to the CSVs for the class names.
Add a onRestore() function to translate the state$input$tissue and state$input$cell.class values (if that's possible).
When single cluster selection is checked, the "rest of region" comparison cluster can be chosen as one of the options.
When meta-groups are enabled, the rest of region option is not available.
In both cases, it would be convenient to have a "rest of cluster" option for the comparison cluster when in the subcluster mode.
The rest of cluster could be computed dynamically, but the rest of region may be too computationally intensive and/or the number of possible comparisons is too high.
a4galt is an example where the tSNE plot fails. It works when an individual region is chosen, so I'm guessing it's because one of the regions has no expression data for that gene.
"Ento Peduncular" should be "Entopeduncular"
For a new user, after a short period slide out a video viewer with a "get started" video that shows in, maybe 30 seconds, what can be done. Besides the play, include two buttons, "Don't show again" and "Remind me later".
If a user enables the "Use common names for subclusters" flag after a filter query is applied, then results are wonky: table is blank, tSNE raises error.
Example: http://dropviz.org/?_state_id_=1bedbda57f9eb018 then change flag.
Example: HC Cluster 11.
Shows "No ICs" in top panel, but table below display 2 real ICs. Selecting an IC raises an error in the tSNE plot generation.
I'd like to post on a DropViz a detailed protocol for making the brain suspensions that we used to build the atlas.
Can you please put it up on the "Data" Section?
I think having a header (in addition to "Metacells" & "Annotations") that says "Single Cell Suspension Protocol from Acute Adult Brain"
and then have the name of the file, which when clicked on, will download the PDF
Saunders_scBrainSuspensionProtocol_v1_180419.pdf
full path:
/Volumes/broad_mccarroll/arpy/manuscript/atlas/draft/data_share/files/Saunders_scBrainSuspensionProtocol_v1_180419.pdf
A light grey "No data" label displays in the center of FC 6 subclusters.
Create a startup script for a GCE instance group to automatically bring up new instances
when I wanted creat a seurat object after loading dge file,I get the error:
Error in CreateAssayObject(counts = counts, min.cells = min.cells, min.features = min.features, :
No cell names (colnames) names present in the input matrix
the martrix read in R don't have the dimnames(rownames and colnames),what should I do?OR How can I get it?
http://dropviz.org/?_state_id_=0065fae0b4dff4d1
If cluster is removed, then success. But apparently no subclusters associated with this cluster?
Remove less recently used files from www/cache/.png and www/cache/metacells/, not www/cache/ic.
This could just be a system call that is part of a one-time startup event. Perhaps if it's scheduled in the future then there is no impact on user experience?
Example: http://dropviz.org/?_state_id_=31f4b6e2eedbe79b
Either nothing displays or IC table doesn't display.
When a single cluster is selected, then the range of the tSNE plot is reduced, but should stay the full size of the entire set.
http://dropviz.org/?_state_id_=06980750fda8f838
Not sure why this is failing. Min/Max should be set to the bounds of all the bags. Or maybe I should pre-compute the range for all tSNE?
Line 10 in 2629965
Add a button that sets a cookie for display preferences. Load the display preferences on new sessions unless restoring from a bookmark.
http://dropviz.org/?_state_id_=8320557b8fc61611 shows individual cell expression.
When changed to Overlay, additional cells are shown associated with the major cluster.
http://dropviz.org/?_state_id_=e9075b6985280802
What are these extra cells? Is there a bug that is not showing them without Overlay? Or is there a bug that is adding cells in Overlay that shouldn't be displayed?
Describe the data sets available on the site by adding structured data for crawler:
https://developers.google.com/search/docs/data-types/dataset
Endothelial_Tip -> Fibroblast-Like
Endothelial_Stalk -> Endothelial
Sometimes, but not always, the inputs in the filter-params div do not reset upon resetQuery action.
Example is first link from manuscript.
I've tried reordering, using shinyjs's reset and directly updating the selected options to "". But nothing fixes the problem.
If a user clicks in the select box, then options are displayed. Even if nothing is then selected, then triggering resetQuery works. This suggests to me that there's something wonky with some inconsistent state in the selectize object when shiny restores from a bookmark.
Scatter plots are not getting restored from bookmarks because the current.cluster and comparison.cluster are not getting set. They are stored as integers referencing, IIRC, a unique ID for that cluster.
This is apparently not a legacy issue, but happens when bookmarking new scatter plots, too.
Where did that come from?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.