plotly / dash-bio Goto Github PK
View Code? Open in Web Editor NEWOpen-source bioinformatics components for Dash
Home Page: https://dash-gallery.plotly.host/Portal/?search=Bioinformatics
License: MIT License
Open-source bioinformatics components for Dash
Home Page: https://dash-gallery.plotly.host/Portal/?search=Bioinformatics
License: MIT License
Hi @matthewchan15,
I'd love to go over how you set up the Heroku deployment via GitHub, which is triggered every time one pushes to a feature branch.
As discussed with @VeraZab, it would be great to automate this deployment also every time the author of a pull request merges into master.
/cc @shammamah
Hi @matthewchan15 @shammamah @VeraZab!
When visiting the gallery and hovering for each component's description, Circos's reads very different from all others:
This would help sort out which ones are needed and which ones we could erase as well
List all stylistic or functional issues here.
When done, change issue assignment to the component creator for them to fix the issues.
When the user selects a custom
Circos graph, and than selects a data set
, Dash-tables
throws an error, that makes the application unusable.
as per: #28 (comment)
@matthewchan15 When syncing up my master
after you merged PR #89 (which came with commit 7154cb2), I noticed that file dash_bio/bundle.js
got huge... Was everything you added necessary?
For reference:
$ git pull origin master
remote: Enumerating objects: 10, done.
remote: Counting objects: 100% (10/10), done.
remote: Compressing objects: 100% (4/4), done.
remote: Total 10 (delta 6), reused 9 (delta 6), pack-reused 0
Unpacking objects: 100% (10/10), done.
From github.com:plotly/dash-bio
* branch master -> FETCH_HEAD
87bd4a0..66ca418 master -> origin/master
Updating 87bd4a0..66ca418
Fast-forward
dash_bio-0.0.1.tar.gz | Bin 1081660 -> 2086631 bytes
dash_bio/Circos.py | 8 +-
dash_bio/DashCircos.py | 83 --
dash_bio/DashIdeogram.py | 194 ---
dash_bio/bundle.js | 7977 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--
dash_bio/metadata.json | 14 +
dash_bio/package.json | 2 +-
package-lock.json | 175 +--
package.json | 2 +-
src/lib/components/Circos.react.js | 11 +-
tests/dash/app_circos.py | 2 +-
11 files changed, 7969 insertions(+), 499 deletions(-)
delete mode 100644 dash_bio/DashCircos.py
delete mode 100644 dash_bio/DashIdeogram.py
This has come up a few times. I develop in Python 3.7, so it's wrong...
I noticed that generated files in dash-bio/dash_bio (such as bundle.js etc.), or the .tar.gz, are versioned in this repository. I wonder why this is the case: usually (at least in my experience) files which can be generated are not integrated in the version control, in particular for large files.
List all stylistic or functional issues here.
When done, change issue assignment to the component creator for them to fix the issues.
Really just a suggestion, but I believe it would be nice to have setup.py require dash by default:
Lines 11 to 20 in bd5ce53
That way, when the user installs dash-bio, they simply need to create a venv and run:
pip install dash-bio
with dash, dash-html-components, dash_renderer being installed as well.
It could be implemented along the line of:
setup(
name=package_name,
version=package["version"],
author=package['author'],
packages=[package_name, '{}/utils'.format(package_name), '{}/component_factory'.format(package_name)],
include_package_data=True,
license=package['license'],
description=package['description'] if 'description' in package else package_name,
install_requires=[
'dash', 'dash-html-components', 'dash_renderer'
]
)
The files in there will need to be cleaned as well
I have seen that this header is part of a few PRs:
html.Div(
id='header',
className='banner',
children=[
html.H2('Dash bio: Manhattan plot'),
html.Img(
src='assets/dashbio_logo.svg',
style={
'height': '100',
'float': 'right',
}
)
],
style={
'width': '100%',
'display': 'flex',
'flexDirection': 'row',
'alignItems': 'center',
'justifyContent': 'space-between',
'background': '#A2B1C6',
'color': '#506784'
}
)
So imho it makes sense to move this bit into a common component, s.t. future changes are easy?
Though this could also be an entire page component wrapper, s.t. e.g. a footer for all components can be added later easily.
I created a self contained demo of the reduced problem at https://dash-gallery.plotly.host/dash-volcano-bug-app/ (the repo is at https://dash-gallery.plotly.host/GIT/dash-volcano-bug-app).
When you use the dropdown to select Set2 from Set3 (which are the same), it is fine. The problem arises when you select Set1 and then again Set2 or Set3, then a whole bunch of data point are not rendered but you can see that they are there with the hover info... you can get the whole dataset to be displayed properly by setting the Thershold to 7 and then back to its initial value of 4, then everything works properly until one selects Set1 and then Set2 again...
I have checked that the data sets that were sent to the figure
prop of the dcc.Graph
were the one I expected and they were. For some reason they do not get rendered.
List all stylistic or functional issues here.
When done, change issue assignment to the component creator for them to fix the issues.
List all stylistic or functional issues here.
When done, change issue assignment to the component creator for them to fix the issues.
ESlint raises the error "'view.length' is missing in props validation (react/prop-types)" on line 118
List all stylistic or functional issues here.
When done, change issue assignment to the component creator for them to fix the issues.
Now we have integration testing set up, we need a common test suite that is used for all common features and files specific to each component's features.
@Bachibouzouk has already started on his (#102) for Volcano plot.
We still need to finish tests for:
First-level impressions clicking through Dash Bio gallery beta apps: https://dash-gallery.plotly.host/dash-bio
Would be great to have a GitHub link in the upper-right of the header that links to the code for each app on GitHub.
mtcars
dataset is a "classic:"https://cran.rstudio.com/web/packages/heatmaply/vignettes/heatmaply.html
closest
by defaultStem thickness
are too large in font-size IMO.hover mode
to closest
The rest of our components don't have the "Dash" prefix, so it would look better if these components didn't have it either :)
Goals before releasing dash-bio:
Circos: https://dash-gallery.plotly.host/dash-bio/circos
https://cran.r-project.org/package=RCircos
https://cran.r-project.org/web/packages/SanzCircos/index.html
https://github.com/hpeaker/ggcircos,
https://github.com/lvulliard/BioCircos.R
Ideogram: https://dash-gallery.plotly.host/dash-bio/ideogram
https://bioconductor.org/packages/release/bioc/html/ggbio.html
https://github.com/freestatman/ideogRam
Clustergram: https://dash-gallery.plotly.host/dash-bio/clustergram
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5925766/
https://cran.r-project.org/web/packages/shinyHeatmaply/index.html
Manhattan: https://dash-gallery.plotly.host/dash-bio/manhattan-plot
http://sahirbhatnagar.com/manhattanly/
Volcano: https://dash-gallery.plotly.host/dash-bio/volcano-plot
http://sahirbhatnagar.com/manhattanly/
Needle plot: https://dash-gallery.plotly.host/dash-bio/needle-plot
https://github.com/freezecoder/mutsneedle
Cytoscape: https://dash-gallery.plotly.host/cytoscape-phylogeny
https://saezlab.github.io/cyrface/
https://bioconductor.org/packages/release/bioc/html/RCyjs.html
3d molecule viewer:
http://bio3d.ucsd.edu/pca-app/?SSUID=2018-10-30_180238089b05
http://thegrantlab.org/bio3d/index.php
MSA viewer:
https://zachcp.github.io/msaR/
http://resources.qiagenbioinformatics.com/manuals/clcsequenceviewer/current/index.php?manual=Sequence_alignment.html
Phylocanvas
https://thibautjombart.github.io/treespace/
https://github.com/zachcp/phylocanvas
https://zachcp.github.io/phylocanvas/
Oncoprint
https://github.com/dakl/oncoprint
https://www.bioconductor.org/packages/release/bioc/vignettes/ComplexHeatmap/inst/doc/s8.oncoprint.html
Gene track/coverage plots
http://www.sthda.com/english/wiki/ggbio-visualize-genomic-data
https://bioconductor.org/packages/release/bioc/html/ggbio.html
https://bioconductor.org/packages/release/bioc/vignettes/seqplots/inst/doc/SeqPlotsGUI.html
Speck
Running python tests/dashbio_demos/app_speck.py
works and runs the app correctly, but when trying to run it with gunicorn, which will happen on the server (gunicorn tests.dashbio_demos.app_speck:server
) it fails with the following error:
Failed to find application object 'server' in 'tests.dashbio_demos.app_speck'
We need to expose the server variable somehow within the app file. I've tried declaring server = app.server
within run_standalone_app
and returning it, but that doesn't work either.
@mkcor @Bachibouzouk Any ideas on how to get this to work? We might have to start running the apps from index.py
after all.
I guess this will happen too in dds
I think we should sort the apps alphabetically and make links to apps display alphabetically.
The syntax in https://github.com/plotly/dash-bio/blob/master/dash_bio/component_factory/_manhattan.py#L440 and https://github.com/plotly/dash-bio/blob/master/dash_bio/component_factory/_manhattan.py#L453 may need to be updated.
when running locally (after pip install -r requirements.txt
) I'm getting:
(venv3) โ dash-bio git:(master) python index.py
Traceback (most recent call last):
File "index.py", line 31, in <module>
for filename in appList
File "index.py", line 32, in <dictcomp>
if filename.startswith("app_") and filename.endswith(".py")
File "/Users/chelsea/Repos/dash-repos/gallery-apps/dash-bio/tests/dash/app_manhattan_plot.py", line 12, in <module>
fig = dash_bio.ManhattanPlot(df) # Feed the data to a function which creates a Manhattan Plot figure
File "/Users/chelsea/Repos/dash-repos/gallery-apps/dash-bio/dash_bio/component_factory/_manhattan.py", line 165, in ManhattanPlot
highlight_color=highlight_color
File "/Users/chelsea/Repos/dash-repos/gallery-apps/dash-bio/dash_bio/component_factory/_manhattan.py", line 440, in figure
suggestiveline = go.layout.Shape(
AttributeError: module 'plotly.graph_objs' has no attribute 'layout'
as mentioned in : https://github.com/plotly/streambed/issues/11788#issuecomment-436746312
developers
/dataliberation
user on ddsdevelopers
as userFrom @Bachibouzouk's comments on #70
_clustergram.py
(fixed in #77)Should help to close #93. WIP in #110.
This could be applied to other files across the repo as well. Preferably once the tests are written so we can use the tests to check that the code is not broken :)
I propose to change name of tests/dash
package to tests/dash_demo_apps
dash-bio/tests/test_volcano_plot.py
Line 76 in 84bb00e
We already have APP_NAME
as a variable, so we can use that in the URL string in this comparison.
Also, the URL would be valid even with an extra /
at the end -- so the test might fail incorrectly
List all stylistic or functional issues here.
When done, change issue assignment to the component creator for them to fix the issues.
List all stylistic or functional issues here.
When done, change issue assignment to the component creator for them to fix the issues.
Hi @matthewchan15,
I'm not a biologist (@shammamah can you please remind me who are the biologists so I can tag them?) but under "Organism" (species?) I find it odd to see "Human" and "Homo-sapiens" (also, I believe you don't hyphenate it).
Now I'm aware they're pointing to two different datasets, which contain different features (data-wise). If we want to include both datasets, we probably want to figure out something UX-wise so it gives a better first impression.
PS: This follows from a PR #90 I originally opened and then closed.
The data
folder here needs to be moved into tests/dash/sample_data
: https://github.com/plotly/dash-bio/tree/master/.github
I thought this might be a plotly.js issue, but I don't get this off synch behavior in Chart Studio: https://plot.ly/~jackp/18501/#/
So I wonder if this app is using an older version of plotly.js.
The JS console is full of these bundle.js errors from click-dragging. I see calcdata
in the compressed source, so it does seem plotly.js is having trouble in this particular environment.
As mentioned in #110.
once this is merged: #21
Hi @shammamah,
As I was linting app_clustergram.py, I noticed some heavy and confusing wording, lines 162-164:
'Header of row labels column in uploaded dataset',
title='If a dataset was uploaded, enter the header of' +
'the column that contains the title of each row.',
Even when I view it in context, it's unclear what this label means (I mean 'label' in the context of an app layout).
In Pandas language, when talking about a DataFrame, row labels are referred to as the index and column labels as columns.
Reference: https://pandas.pydata.org/pandas-docs/stable/dsintro.html
Maybe we could follow this convention, instead of torturing our brains?
Ideogram
https://github.com/igvteam/igv.js/tree/master
https://igv.org/web/test/examples/bam.html
Clustegram
http://qb.cshl.edu/ginkgo/?q=results/_t10breast_navin
https://github.com/robertaboukhalil/ginkgo
Sequence viewer
https://hemoglobin-comparison.herokuapp.com/
https://github.com/lmalet/hemoglobin_dash_app
Will be cool to eventually replace this graph with 3d molecule viewer or Speck component
In accordance with https://github.com/plotly/streambed/issues/12263, we are making our index.py
app obsolete. The applications we currently have are not true standalone apps, since they just modify the layout on the main index.py
app. Right now, index.py
lives on that server (https://dash-gallery.plotly.host/dash-bio) and its portal (https://dash-bio.plotly.host/Portal/)
What we need to do:
requirements.txt
file, add in the dash_bio
dependencydemo_apps
that contains a folder for each appsample_data
files into its folderconfig.py
and runtime.txt
) for each app in its folderLet's get this done before I fly out to Boston for training with Amgen (13 Feb)! They're a huge pharmaceutical company and it would be awesome to show off Dash Bio and sell them on DDS simultaneously :) let's get that ๐
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ 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. ๐๐๐
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 โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.