You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
From the publication and vignette here, it's unclear what is the expected performance of the model. For instance, in BasicExample.ipynb your model assigns weight to each explanatory variable. In the publication, the top 100 genes are selected for further analysis (how? top betas? What parameters yielded those results?).
Some clearer explanations and examples on how you would recommend tuning parameters and training the model would be helpful to include in this repo. For scientific purposes, it would also help in reproducing results from the publication.
Hi, we selected the top 100 genes based on the absolute values of betas.
In our paper, we noticed that tuning the parameters according to the desired number of SNPs is a good strategy.
This repository only serves as an example for people who are interested to use it, if you want to replicate the results for scientific purpose, I'm happy to share the private repo with all the scripts with you. (I will edit the README to avoid further confusion of others about this point.)
Activity
HaohanWang commentedon Jun 18, 2019
Hi, we selected the top 100 genes based on the absolute values of betas.
In our paper, we noticed that tuning the parameters according to the desired number of SNPs is a good strategy.
This repository only serves as an example for people who are interested to use it, if you want to replicate the results for scientific purpose, I'm happy to share the private repo with all the scripts with you. (I will edit the README to avoid further confusion of others about this point.)