multi-omics tutorial

[dorothyzh]

For the tutorial, only shows the single omics analysis using Stabl, but the cyto_ool and other datasets were not used in the tutorial. So could I find the tutorial or the codes for multi-omics elsewhere?

[xavdurand]

Hello @dorothyzh ,

Thank you for your comment. We added a quick explanation of the 3 ways of handling multi-omic datas: https://github.com/gregbellan/Stabl/blob/main/Notebook%20examples/Handling%20Multi-omics.docx

I closed the issue with this file.

[dorothyzh]

Thanks, but if I would like to use the third way, which function should I use for stable? And for multi_omic_stabl_cv function, the plotting always have some bugs inside, could you provide a multi-omics tutorial similar as the "STABL in single-omic" one in the tutorial, then we could directly get the stabl object and omit plotting issues.

[xavdurand]

Thank you @dorothyzh for your comment. I advise you to look for the scripts named run_*.py for an example of run.

Is it possible for you to describe the issues you encountered? You can use this template:

---

Describe the bug

A clear and concise description of what the bug is.

How To Reproduce

Describe steps and provide code if possible

Expected behavior

A clear and concise description of what you expected to happen.

Priority

What is the impact of this bug on the user, how critical is to fix? P0, P1 .. P4 Reference - Google Issue tracker priority levels

Desktop

OS: [e.g. MacOS, Linux, Windows] Python Version [e.g. 3.9, 3.10, 3.11, 3.12] Branch [e.g. main, stabl_lw] Other relevant information

Additional context

Add any other context about the problem here.

[dorothyzh]

Thanks for the comment, and may I ask for some detailed description of the results, for instance, is there any more detailed results other than feature names that we can get from the object "stabl_regression". Such as the predicted values or p value or any statistics for these features?

[xavdurand]

Hi @dorothyzh ,

You have multiple attributes of the Stabl object that you can use. For example, the parameter stabl_scores_ contains the statistical score of each feature calculated by Stabl. These scores are not properly p-values but frequencies of selection. You can take a look here for more attributes: https://github.com/gregbellan/Stabl/blob/30039a573d072c4604396197466768951458ca3f/stabl/stabl.py#L807C3-L838C69

Stabl's main purpose is the feature selection so a predicted value is not meaningful, as it implements SelectorMixin scikit-learn object. You can use Stabl as a selector/transformer before the modelization step.

[xavdurand]

As the issue seems resolved. I close the github issue.