molecularcartography / ili Goto Github PK

Currently the user could be notified when something goes wrong in file parsing (csv, stl, png) but it works inconsistently. If file has completely wrong format there is no notification. It's confusing (like in #11).

see details bf2e367

to be able to zoom in into different regions of the model

Jeramie example, map 997.1115

changing the hs-quantile from 1 to any other value (e.g. 0.99) makes most of spots (having the same intensity) transparent, whereas it should change only the colors of spots with highest intensities

Error dialog: Operation failed. See log for details
Log:

I just wanted to +1 this request
perhaps a general solution would be an icon that lets a user switch between rotate and pan

Ctrl+up/down interfere with system-wide shortcuts and don't work. Ctrl+S/O is unusual on mac.

hot colour map should have dark red for minimum value going up to white

You can change the values being displayed by clicking on the bottom right of the screen. This is not intuitive for common users - at least it wasn't for me.

To set values corresponding to the first/last colors in the colormap.

Use-case: render two (or more) maps with comparable colors
1. have a look at two maps, write down min (m) and max (M) intensities across both of them
2. set values of the first/last color from auto to the m and M
3. plot each map separately, export each map into a PNG file
4. switch the mode into "auto"

csv is fine for most purposes, but some of my category labels have commas, which I have to change to periods or something else. Support for tsv files would be a convenient solution.

it's better to use a free one than a proprietary one

viridis: https://gist.github.com/kanchukaitis/bcf0f7a3d1556cbfdb05
short explanation: http://bids.github.io/colormap/
talk on it: https://www.youtube.com/watch?v=xAoljeRJ3lU

Currently the user needs to open MapSelector to change map. Since it's expected to be important operation we want to let changing a map from any context.

In addition to up/down/page-up/page-down key strokes to handle Ctrl+* strokes. Ctrl variant differ in the following way: it works in global context (whatever focus in, except opened combobox - platform limitation) and and if MapSelector isn't active it shows it shortly with 0.1 opacity (to give the user sense of conntext).

Ctrl+page up/down is intercepted by browser and won't work in a web page.

Some explanation regarding showing the MapSelector. Result of up/down depends on current filter. If we ignore the filter on Ctrl-... it would be confusing. If we apply the filter now showing it's applied it also may be confusing.
If we handle up/down globally (without Ctrl) we may interfere with dat.gui and cause the same sort of confusing that OrbitController caused with lighting. Ctrl gives some sort of uniformity - it's how we handle context independent shortcuts.

This is admittedly a minor issue, but forward slashes occur in some of my category labels and don't display properly in the drop menu.

Currently, there is no easy documentation or tool about how to place the 3D points (where the sample was taken). This info/tool will improve usability for non experts.

is updated only when afterwards dragging the model.

maybe event listener for this slider doesn't work?

...(if not too complicated)

It would be really nice to have a checkbox to flip the color map. For example, RdBu goes red-white-blue, but if I want high values to be red and low values blue, I need a color map that goes blue-white-red. Flipping the color map, or adding a BuRd color map if one exists, would allow for this.

Currently, there is no easy way to create models. The only options are to use a 3D scanner or buy a model. One option is to have a stl library with common models and perhaps a tutorial about how to modify or create your models.

The issue is created to check the synchronization between GitHub and Asana using Zapier.

...with another slider of intensity “Front light” in “3D” dat.gui

the front light is the light with the coordinates equal to the coordinates of the camera

for multiple views, either: multiple lights, or the light only for the first view (what's easier)

example: min/max = 1/10
log(min/max) = 0.1/1.3

And there is no way to return back to `ili but loading model/intensities again and set the settings

Possible solutions:

• open http link in a new tab/window
• warning before leaving `ili
• proper Back

either with no intensity values given or log(0)

The issue is created to check the synchronization between GitHub and Asana using Zapier.

probably if the name contains "." already e.g. 242.0124 in the rapidfire example or 253.12 in the bee full example

At this moment you can only do a print screen (mac cmd-s). It will be great to generate high res images, like svg or pdf. The latter should be preferred.

It seems like it doesn't consider the x-y-z adjustments

This is how it looks at the screenshot:

This is how it looks in the Ctrl-s'aved PNG:

currently it replaces the current window so the whole map has to be reloaded.

minor as immediate workaround is to right click and 'open in new tab/window'

Design doc: https://docs.google.com/document/d/10AJ-x3GakM8ExQ8ZDfHPhYAxD2zStQjbpdNtOJv4jxg/edit?usp=sharing

User should be able to zoom to a point of interest.

Now scaling with mouse wheel preserves point at the center of the screen.

When new image or mesh loaded appropriate view should take appropriate zoom factor for current layout and image/mesh size.

It would be nice if separate molecular maps could be displayed simultaneously on different views in 3D mode. It would allow visualizing correlations between the maps.

rather than in a separate window

as discussed: using the previous way of creating an App (1.0?)

The issue is created to check the synchronization between GitHub and Asana using Zapier.

Can you please slow it down by at least 2 times (if it doesn't make it unreasonably slow on Win/Lin)?

with radius offset along x, y, and z i.e. at (x+r,y+r,z+r). Some labels will not be visible (hidden by the model), but this is fine

User should be able to remove the labels completely or adjust font size and color.

Currently, in the same file you need to have the 3D value point and the values to plot in the model. I think a better way to go about this is to have:

• The model (stl)
• The sample names and position in 3D (csv) -- we could even perhaps merge this and the previous one
• Multiple input formats where you have different information (rows vs features).

I believe this will be more convenient because the first 2 only are needed once but the 3rd kind can hold different kinds of values and could constantly change: family or species level taxonomic information, metobolomics, PCoA axes, etc.

specifying http://...csv and http://...stl as arguments to the toolbox URL
STEP1: on the same server (inside the ili folder)
STEP2: on the external website

Design doc: https://docs.google.com/document/d/1cZxk0jqCJEsxZgWHLIrvmDQ36WuTnqhmW76ykVREkng/edit?usp=sharing

Can a mouse-over be easily implemented that shows the label of data points and provides a html link to another webpage?

perhaps this could be implemented with html code embedded in the data labels

files: lung.csv/lung.png from demo
system: OSX 10, chrome Version 42.0.2311.135 (64-bit)

to reproduce: drag-drop pair of files into browser
result: data fails to load, see screenshot for log

dataset (15K spots) in the db, example_2D_Jeramie/bugreport. Ctrl-s of any map eats 1GB memory, second Ctrl-s (either the same or another map) eats 7GB memory, third Ctrl-s normally has not enough memory and freezes the webpage.

using a smaller PNG (e.g. 50% x, 50% y) doesn't help

...rather saving always with the black background