achtman-lab / grapetree Goto Github PK

All profiling methods fail from running in App (issue with launching subprocess)?

MSTree fails when running with

python MSTree.py profile=examples/simulated_data.profile method=MSTree edge_weight=eBurst

TypeError: apply_along_axis() got unexpected keyword argument 'minlength'

Comma delimited metadata does not work. Try sample files attached

microreact-project-NyqrhlsB-tree.nwk.txt

microreact-project-NyqrhlsB-data.txt

rMLST.rep.MSTv2.nwk.txt

Maybe an issue with Labels.

select all
unselect all
show/hide metadata
show/hide hypothetical nodes

If there is no tree loaded, pressing buttons like save tree, or centre graph, the browser throws a big ugly error and buttons dont do anything. Either users should not be allowed to press those buttons; or there should be message

Allows to specify the location, size, font size, colors, number of showing categories, ...

In NJ tree, when I select only hypothetical nodes and click "Load selected", I should get a warning that explains why I am being stupid.

Javascript can handle it, could be usefu having it read excel sheets.

see: https://aymkdn.github.io/ExcelPlus/

Show hypotheical nodes, such that the users can select and move them.
Then disable them for a pretty figure.

The splash screen is very informative, but once I've loaded a file for the first time and want to load another one, i cant find it. Either there should be a button ("Help") that brings it back, or it should up when an exception/error is thrown.

Fixed in latest commit,

There should be different metrics for collapsing branch. Either on an absolute cutoff (what it does at the moment) Or is should measure the branch distance from tip to root and do so cumulatively, This will allow people to just prune the tips without losing the deeper structure.

Better exception handling for BASA/ maketree step: When this fails I get a singular error message back with no consideration for the exception thrown at the back end. We may want to have more informative message cause its very hard to troubleshoot (even for me).

BLACK balls show up with too many categories, there should just be an upper limit when you run out of colours. It's not meaningful showing 300 different groups on a tree. Also, the default column it loads from the metadata should NEVER be the ID.

When I have loaded a profile, and change the parameters, I pressed refresh and nothing happened. The only way to change these parameters and affect the tree is at load time apparently. Either these options should only be set at load time, or the refresh button should recalculate the tree with those new settings.

In the context menu (last item in the menu), when clicking on "Figure legend", the options run out of the window and I cannot scroll to reach them.

40 is mislabelled as unspecified.

• Separate strains duplicate legend and then it doesnt go away.
• svg output font is weird serif font.

• Embedded help, especially for key combinations to do things
• Help link for metadata & Main window
• Help links on contextual menus.

The menu buttons under "Enterobase" are a bit confusing.
Suggestion:

• remove "Highlight selected" and move this over to the Enterobase window
• "Load selected" and "Filter selected" are kind of the same, keep only load
• Synch -> Sync ( can be a checkbox?)

Then there are 4 buttons left:
"Load selected nodes"
"Import fields"
"Save tree"
"Sync" <- can be a checkbox

There should be some notification for the user that all these buttons trigger a change in the enterobase table based on selections in the GrapeTree.

the initial placement of the figure legend overlaps the menu on the left hand side

• laptop screen
• firefox

suggestion: figure legend initially placed on right side of the window looks nicer

Memory error on edmonds for Windows machine using fairly large allele profile.

rMLST.rep.txt

Traceback (most recent call last): File "MSTrees.py", line 340, in <module> tre = backend(**dict([p.split('=') for p in sys.argv[1:]])) File "MSTrees.py", line 330, in backend tre = eval('methods.' + params['method'])(names, profiles, **params) File "MSTrees.py", line 216, in MSTree tree = eval('methods._'+matrix_type)(wdist, **params) File "MSTrees.py", line 89, in _asymmetric mstree = Popen([params['edmonds_' + platform.system()]], stdin=PIPE, stdout=PIPE).communicate(input='\n'.join(['\t'.join([str(dd) for dd in d]) for d in dist.tolist()]))[0] MemoryError

Seems to run to completion in linux version

would be good

Should block over 8K nodes and should explain how to simplify dataset.

How to Search for related strains. For cgmlst only.

the branch scaling slider/ option causes two issues in automatic rendering mode:

1. the scaling is only triggered after I click on something else, it is not scaling with the slider
2. after adjusting branch lengths, the tree is frozen and nodes cannot be moved (it can be fixed when switching to static rendering and back again)

"tree construction is not available" message, i still see the dropdown for the method select (NJ, BASA, MST), should be hidden

Need to remove and migrate functionality to load tree button.

• menu header Inputs/Outputs -> Exports
• button "unmodified tree" is broken/ should be removed?

Outputs -> Export

Label text right after Show labels and possibly remove title.

Relative scaling -> Kurtosis

add metadata -> add column

search in label -> highlight label

If one of the taxa in the profile has a ' : ' it in, this will break the final newick file, as this is a special character. As such, mstree.py needs to sanitize the input labels for " : ( ) , "

Also the newick file loaded doesnt seem to ever stop loading and recognise there is something wrong with the file.

In the metadata table, the headers are slightly shifted which gets confusing when I scroll through the table.

Hypothetical nodes seem to be shown in the legend even when it is set not to be shown (default)

When loading metadata, having the ID column is very specific, it shoudl be case insensitive. ID only at the moment, but it should allow id. Rather than being so strict, If the ID column specification fails, the program should just try to use the first column. If That also fails, then give a message that ID cannot be found (at the moment it just says my file is no good but doesnt give any hints as to why).

Allows users to specify the parameter by a input box.

Change the drag bars into log scale.

1. multi column legend
2. allow assignment of a new column to UDF after created.
3. Multi column sorting in metadata grid
4. SNPs -> chose presentation between GrapeTree and phylogram.

Dont make the file name in the top left shown in italics, it too hard to read. maybe bold?

The selection of the layout algorithm, should be under layout no?

Reset all the parameters to the default values and refresh the page

Feeding a bad file e.g. profile does not have a '#' at the start. will give no feedback (error message).