zakandrewking/escher

Build, share, and embed visualizations of metabolic pathways.

escher's Issues

Implement on_load callback in metabolic map.

To avoid setTimeout hack in the knockout.html example

error building javascript

➜  escher git:(master) ✗ python setup.py build         

Tracing dependencies for: ../build/almond
Error: ENOENT, no such file or directory '/home/aebrahim/escher/escher/js/src/Input.js'
In module tree:
    main
      Builder

Error: Error: ENOENT, no such file or directory '/home/aebrahim/escher/escher/js/src/Input.js'
In module tree:
    main
      Builder

    at Object.fs.openSync (fs.js:427:18)


Tracing dependencies for: ../build/almond
Error: ENOENT, no such file or directory '/home/aebrahim/escher/escher/js/src/Input.js'
In module tree:
    main
      Builder

Error: Error: ENOENT, no such file or directory '/home/aebrahim/escher/escher/js/src/Input.js'
In module tree:
    main
      Builder

    at Object.fs.openSync (fs.js:427:18)

Cleaner callback in knockout.js

There should be a way to return a value from get_flux() in knockout.js. Then, we can remove one extra layer of callback.

reaction labels don't delete

server cannot run from the server directory

Currently this works

python server/tornado_main.py

This doesn't

cd server
python tornado_main.py

Pointed out by @jlerman44.

Keyboard shortcuts occasionally stop working

This happens in the current dev branch.

Support csv upload of flux and node data

@steve-federowicz pointed out that the general public does not have access to json serialization of data (basically, if Excel doesn't have it, don't count on it).

So we should support CSV files for uploading reaction and metabolite data.

consolidate files

Ideally the python module would use other locations instead of its own "static" directory to avoid duplication and needing to update code in 2 places.

Should it also place things in the appropriate directories i.e. its templates in the template directory?

@zakandrewking If you think this is a good idea I can make a PR to tweak the install script and that all that jazz.

example doesn't work because of a dead link?

Hi, the example from http://cobrapy.readthedocs.org/en/latest/visbio.html doesn't work. I get the following error:
ValueError: No map named e-coli-core found in cache or at http://zakandrewking.github.io/visbio/

missing space chars recognition in builder text mode

Space characters inserted in text mode of the builder are not remembered if I save a json map and reload it again.
I tried it online with the last version (1).

Undo/redo behavior is occasionally broken

One instance: Add reaction with one product, cycle metabolite (c) twice, undo (C-z).

feature suggestion

would be great if you could use the map builder w/o having to specify a simpheny map to start with. maybe just a cobrapy structure if all that is needed?

Bidirectional arrows with flux map

The map makes it difficult to tell which direction flux is flowing for reversible reactions.

I'm not sure if this is implemented and not working, or if this is a feature request.

General maps

As I am working with a wide range of bacterial organisms (and not ecoli) I was wondering how to couple known informations such as enzymatic commission numbers to a visualisation perspective? Also I looked for simpheny maps but it seems a closed source private company?

Move to Sphinx + Read the Docs for documentation

See https://github.com/batiste/sprite.js for a nice example.

metabolic-map: BIGG map membranes do not render correctly.

Fill for membranes is black.

Installing visbio to site-packages breaks map-cache

Map-cache cannot run in site-packages because permission usually require root.

Map id's use _ instead of -

Just a suggestion...

Using _ instead of - in id's will make future tab-completion in languages like python easier if that ever were to happen in the future, and these id's will likely be easier to change sooner instead of later.

Implement settings.ini for server

The server has some options that should be set in a settings.ini file. For instance, the location of a cobra model.

svg export from python

Is it possible to export the map made by escher directly from python?
I mean something like this:

b = escher.Builder (...)
b.export_svg (...)

It would be usefull to integrate escher in other web apps.

Knockout map does not load correctly

The Knockout map does not usually load the wildtype flux distribution. Not sure why.

black boxes in svg

When you do an svg export of a metabolic map (I used simpheny-maps/ijo-central.svg), there are black boxes in the background (as in a previous bug, though the black boxes do not appear in the browser from which the svg is exported).

Cannot load the same file multiple times in load_files

In this function, in js/vis/scaffold.js, trying to load the same file for multiple callbacks results in values of the callbacks object being overwritten.

    function load_files(files_to_load, final_callback) {
        // load multiple files asynchronously
        // Takes a list of objects: { file: a_filename.json, callback: a_callback_fn }
        var i = -1, remaining = files_to_load.length, callbacks = {};
        while (++i < files_to_load.length) {
            var this_file = files_to_load[i].file;
            callbacks[this_file] = files_to_load[i].callback;
            load_the_file(this_file,
                          function(e, d, file) {
                              callbacks[file](e, d);
                              if (!--remaining) final_callback();
                          },
                          files_to_load[i].value);
        }
    };

visualisation

I am trying to make a visualization of a metabolic network using COBRApy and escher (as they suggest in the COBRApy manual). I have an sbml file with all the reactions in my organism.

I already generated my own escher HTML file using following code:

model = cobra.io.read_sbml_model('E:/e.coli.sbml')
modeljson = cobra.io.json.to_json(model)
b = Builder(map_json=modeljson)
b.save_html('E:/e.coli.html')
Opening the html file in my browser(s) just results in an empty screen, although the file itself is not empty.

Does anyone have an idea how I could solve this, or get to the bottom of the problem?

The b.display_in_browser(), also does not work (gives a grey empty screen)

Thanks,

Bart

New reaction requires 2 clicks to switch metabolites

pushing b

typing b when selecting a reaction name activates "show control points (b)" instead of just allowing user to type b as in "FBP"

Features for a functional builder

Goal: Build the E. coli core metabolic map in < 10 minutes.

• Change the direction of a new reaction before adding it. (Arrow next to metabolite can be dragged to determine direction)
• Undo with Ctrl-Z
• Combine metabolites from different reactions
• Export as SVG
• Drag bezier points
• Add new control points
• Change the size of the map border
• Add membranes
• Show main products of a reaction (as a hint) in reaction input menu
• Smart bezier displacement when a node gets dragged
• Show flux on builder reactions
• Snap to grid
• Directional arrows to show (a) reversibility and (b) flux direction
• Drag metabolite and reaction labels
• Add new text labels

Anything else? @jlerman44?

how to create own map

i'm using visbio to visualize cobrapy results from the e. coli model what works fine thanks!
i'm wondering and i'm just a little confused how to bring it to work with other organisms?
i have created a model and have a sbml file. when running cobrapy & visbio only very few reactions were highlighted due to the fact that pathways and reaction names differ.
what is the easiest way to create my own map?
i have no simpheny. can i load sbml data into the builder?
thanks for helping

Complete.ly does not always load

The file complete.ly.1.0.1.js does not always load before the rest of the builder files.

save svg

What is the best way to save a map produced with escher & cobrapy as svg?

When I use escher.Builder(map_name=map).display_in_browser() to get the builder with the map, I can save it as svg but there are guides in it (circles which are used in the builder to move the reaction) and even the arrow indicating the direction of a further reaction.
How can I avoid them?

reaction colors

How can I influence the colours shown in a map for active fluxes?

Currently when I open a map with flux data:

escher.Builder(map_name=map, reaction_data=flux)

all reactions are coloured even non existing or ones with zero fluxes.

CSS loads slowly

This effect all plots that load css then update svg without waiting for the css to be ready, including bar.js, histogram.js, box-and-whiskers.js.

Settings menu does not work in viewer mode

Run server from python package

Run the tornado server from the installed python package.

python -m visbio.server

With a help menu:

python -m visbio.server --help

And extensions:

python -m visbio.server.knockout

Inspired by: http://agschwender.github.io/pilbox/

Dragging in select mode caused unwanted motion in pan+zoom mode

Related to this stack overflow question:

http://stackoverflow.com/questions/15147136/how-to-know-the-current-zoom-level-in-d3-js

Turning zoom on and off is tricky because the d3 zoom behavior remembers pan and zoom even when zoom is off. Thus, one has to record the last scale/translate state then move to that state when entering pan+zoom mode again.

SVG export causes issues with Illustrator

The exported SVG from the builder displays incorrectly in Illustrator CS6, but works fine in Chrome.

Paths with markers only display the markers, and the paths are not created on import. Adding a duplicate path with no marker solves the problem, but it's not an ideal solution.

Rotation is broken

Something goes wrong when you rotate after loading a map.

Rotate then undo is broken

Error:

Uncaught TypeError: undefined is not a function escher.1.0.dev.js:2794

Seen after selecting a large section of the map (Ubiquinone reduction/oxidation) and rotating, and trying to undo.

getdatafiles will not work with static html servers / github.io

Hardcode the filenames for github.io.

In templates, maybe demonstrate the hardcoded data option.

Arrows no longer show reaction direction (when data is loaded)

Hi!

I'm using Escher-1.0.0b2. A few weeks ago, when I started using this version, arrows would show the direction of the flux for reversible reactions:

However the behaviour now (I haven't updated anything nor touched the code) is no longer there. For example:

The only thing that I could think of is of an update in the browser, but then I tried a different browser and arrows still showed the same behaviour (not indicating the flux direction). Any ideas of why this might not be working anymore?

Thanks!
Pablo

Wacky IPython Notebook behavior

1. Map extends beyond container.
2. Detecting scroll events is conflicting with notebook scroll.
3. Need to include the buttons in the notebook, but not the menu.
4. Should ignore key events.

usage cases to support

Installation:

1. pip install escher
2. python setup.py install
3. escher.pth file in site packages, or add to PYTHONPATH

Running:

1. Install with any method
2. Run server with python -m escher.server

• If installed with method 3, you will see development options
  1. Run in Python/IPython terminal
  2. Run in IPython notebook
• Run quick server from the notebook
  1. View and edit maps at http://zakandrewking.github.io/escher
• Same interface as the local server (not the development version)

Development:

1. Install with escher.pth file, or add to PYTHONPATH
2. Serve with python -m escher.server or python server.py

• Escher DOES need to be installed for the server to run.

Testing:

• Testing out-of-box with py.test and jasmine-py

Minified file loads too slowly.

Is it loading asynchronously?

I need to take jquery out of the minified file anyway. And that will probably speed it up. If it is asynchronous, I should supply a callback to the global scope. For example:

window.visbio_loaded(function(visbio) {

});

Maps from SimPheny import do not show flux colors for some reactions.

The reaction AKDGH does not change color or get a flux value assigned.

I believe @jonm4024 also noticed this.

Change map ids to use _ instead of - (gh-pages)

This will effect the maps hosted at zakandrewking.github.io/escher

Undo rotation acts strangely

does not undo the complete rotation

Windows installation errors

The following two errors are seen with a Windows installation of Escher:

Microsoft Windows [Version 6.1.7601]
Copyright (c) 2009 Microsoft Corporation.  All rights reserved.

C:\Users\***\Documents\GitHub\escher>C:\Python27\python setup.py ins
tall
Traceback (most recent call last):
 File "setup.py", line 13, in <module>
   version = __import__('escher').__version__
 File "C:\Users\***\Documents\GitHub\escher\escher\__init__.py", li
ne 3, in <module>
   from plots import (Builder, get_cache_dir, clear_cache, list_cached_maps,
 File "C:\Users\***\Documents\GitHub\escher\escher\plots.py", line
14, in <module>
   from jinja2 import Environment, PackageLoader, Template
 File "C:\Python27\lib\site-packages\jinja2\__init__.py", line 33, in <module>
   from jinja2.environment import Environment, Template
 File "C:\Python27\lib\site-packages\jinja2\environment.py", line 13, in <modul
e>
   from jinja2 import nodes
 File "C:\Python27\lib\site-packages\jinja2\nodes.py", line 18, in <module>
   from jinja2.utils import Markup
 File "C:\Python27\lib\site-packages\jinja2\utils.py", line 520, in <module>
   from markupsafe import Markup, escape, soft_unicode
ImportError: No module named markupsafe

C:\Users\***\Documents\GitHub\escher>C:\Python27\python setup.py tes
t
Traceback (most recent call last):
 File "setup.py", line 27, in <module>
   'out=escher/lib/%s'%escher, 'optimize=none'])
 File "C:\Python27\lib\subprocess.py", line 522, in call
   return Popen(*popenargs, **kwargs).wait()
 File "C:\Python27\lib\subprocess.py", line 709, in __init__
   errread, errwrite)
 File "C:\Python27\lib\subprocess.py", line 957, in _execute_child
   startupinfo)
WindowsError: [Error 193] %1 is not a valid Win32 application

C:\Users\***\Documents\GitHub\escher>

Tiling for large maps

Implement using the d3 tile plugin.

https://github.com/d3/d3-plugins/blob/master/geo/tile/tile.js

Demo:

http://bl.ocks.org/mbostock/5593150

merged metabolites lose track of their connected segments

From an email:

I ran into something strange that didn't happen before:

• Fusing the same metabolite isn't working right.
• One of the 'fused' metabolites can be movable around the map and have no effect on the connecting lines
• Deleting this floating metabolite, however, results in the disappearance of the connector lines
• The remaining metabolite that used to be connected by the lines can no longer be deleted. Undoing is also impossible.

json COBRA model is bigger than it needs to be

Separate out metabolite details into a metabolite list.