Implementation of visualisation tools #121
Conversation
…ingedRuslan-master
…DataFrame object and PathToFile
…ing a TypeError if unsupported type provided
…e - scona.visualisations
|
This is SO COOL @wingedRuslan! Way to go! I'll add some comments now 😸 |
There was a problem hiding this comment.
Awesome work @wingedRuslan!
The main piece of work is to add the error bars so that you're plotting across multiple random networks rather than just one. The other comments are quite small.
I haven't had a full run through the jupyter notebook, but great work! I'll look at that at the next pass 😺
scona/visualisations.py
Outdated
|
|
||
| # save the figure if the location-to-save is provided | ||
| if figure_name: | ||
| fig.savefig(figure_name, bbox_inches=0, dpi=100) |
There was a problem hiding this comment.
If the folder doesn't exist yet, does this work? I feel like it probably needs a os.makedirs check...
There was a problem hiding this comment.
Ohhh, did not check that, I will modify code to handle this case
scona/visualisations.py
Outdated
| (e.g. color =["#06209c","#c1b8b1"]) or you can pass an (r, g, b) tuple, | ||
| where each of r, g, b are in the range [0,1]. Finally, legal html names | ||
| for colors, like "red", "black" and so on are supported. | ||
| labelBar : bool, optional |
There was a problem hiding this comment.
style wise: label_bar rather than labelBar? To fit with the other parameters?
There was a problem hiding this comment.
sure, I should be consistent in naming
scona/visualisations.py
Outdated
| network_measures : dict | ||
| real network measures | ||
| Note: the dict could be obtained from calculate_global_measures() | ||
| rand_network_measures : :dict |
There was a problem hiding this comment.
Does this contain multiple random network measures? Or just one?
The plot needs error bars so that you can see the variance of the random networks.
Here's the original code that calculated the global measures for a bunch of random graphs: https://github.com/KirstieJane/NSPN_WhitakerVertes_PNAS2016/blob/5c9c46caf91768d4cadec2b24078b640f05d3d76/SCRIPTS/make_graphs.py#L274
This is the same issue as above (line 49)
There was a problem hiding this comment.
Just one.
Yeah, I did not know that we need a bunch of random values. I've created functions to accept only 1 random values.
This is the same issue as above (line 49)
Right now plot_network_measures(network_measures, rand_network_measures..) accepts
network_measures : dict
real network measures
rand_network_measures : dict (from 1 random graph)
random network measure values
Should I redesign so that the function accepts rand_network_measures which is a list of some random measures?
Like:
{'assortativity': x1, x2, ... xn
'average_clustering': x1, x2, ... xn
'average_shortest_path_length': x1, x2, ... xn
'efficiency': x1, x2, ... xn
'modularity': x1, x2, ... xn }
And then plot with error bars, right?
There was a problem hiding this comment.
I think it would be ideal to accept the dataframe or dict returned by brain_bundle.report_global_measures() if you can figure that out
…rder to improve the process of saving a figure to file.
There was a problem hiding this comment.
Hi @wingedRuslan this is nice work. Really well done, I do however have two requests:
- I have asked for some (very pedantic) changes over the
real_networkvariable name. - There is a function that you seem to have deleted and incorporated into the place where it is used. I prefer more functions, but I'm happy to be challenged
scona/visualisations.py
Outdated
| from scona.helpers import save_fig | ||
|
|
||
|
|
||
| def plot_rich_club(brain_bundle, real_network, figure_name=None, color=None, |
There was a problem hiding this comment.
It seems like #139 might be resolved
scona/visualisations.py
Outdated
| While instantiating GraphBundle object we pass the real Graph and its name. | ||
| (e.g. bundleGraphs = scn.GraphBundle([H], ['Real Network'])). | ||
| To plot rich club values along with the rich club values from randomed graphs | ||
| it is required to pass the name of the real network (e.g.'Real Network'). |
There was a problem hiding this comment.
I'm quite cautious about the notion of "real graph" that you use here. I would perhaps say
real_network: str, required
This should index the particular network in `brain_bundle` that you want the figure to highlight. A distribution of all the other networks in `brain_bundle` will be rendered for comparison.
I would also prefer to rename real_network to something like distinguished_network or main_network, because the word real raises certain assumptions
scona/visualisations.py
Outdated
| `brain_bundle` will be rendered for comparison. | ||
| To plot rich club values along with the rich club values from | ||
| randomed graphs it is required to pass the name of the initial network, | ||
| the proper network, the one you got from the mri data. |
There was a problem hiding this comment.
I would like to remove the following from all function descriptions:
To plot rich club values along with the rich club values from
randomed graphs it is required to pass the name of the initial network,
the proper network, the one you got from the mri data.
I find small_coefficient to be rather confusing... I think small_world_coefficient makes much more sense.
I find small_coefficient to be rather confusing... I think small_world_coefficient makes much more sense.
…scona into wingedRuslan-master
…ngedRuslan-master
Update forked repo before opening PRs
… on WhitakerLab:master [WIP] Edits to visualisation tutorial
|
Hey @wingedRuslan! This all looks great to me! If you can merge wingedRuslan#6 then I'm totally fine for you to merge! @Islast - you requested changes and I think they're all addressed. Would you mind taking a look and approving when you have a chance? |
Rename tutorial; visualisations_tutorial -> global_measures_viz
|
@KirstieJane, Thank you a lot! I would like to merge but as @Islast requested changes, her approval is needed. |
KirstieJane
dismissed
Islast’s stale review
Chatted with @Islast offline and she's happy to merge
|
Wooooo - click the big green button @wingedRuslan!! MERGE AWAY 🚀:star2: |
(this is a good place to reference any issues that this PR addresses)
Visualisation tools
References issues #116 #115 and #119 #118
I've implemented visualisation tools for the following purposes:
plot_rich_club- for plotting the rich club values per degree along with the random rich club values created from a random network with a preserved degree distribution.Result was before comments
Result now after comments
plot_network_measures- plotting network measures values along with the random network values values created from a random networkResult
plot_degree_dist- plotting the degree distribution along with the degree distribution of an Erdos Renyi random graphResult was before comments
Result now after comments (no gap)
Kirstie's code was really helpful but mostly I've designed these functions by myself, so there are might be some features(changes) you would like to have and on the contrary not to have.
Basically, I tried to follow the Kirstie's code (like to have the same parameters to the functions), but the main implementation was either hard to reproduce or had errors while running. That's why I relied on my intuition and created functions from my perspective.
Feel free to point out me any wrong decisions I took while implementing visualization tools.
I have also tried to have the same outputs illustrated in issue #1.
The way how to use created functionalities I showed in the visualisations_tutorial.
The design of the functions and the produced figures
(e.g the wiki or the WhitakerLab website)
Documentation is updated to include a new module - visualisations.py