search

HAN-Siyu / LION

An Integrated R Package for Effective Prediction of ncRNA- and lncRNA-protein Interaction
https://doi.org/10.1093/bib/bbac420
GNU General Public License v3.0
9 stars 3 forks source link

docs: update package version to v0.2.9 #2

Closed amatria closed 1 year ago

amatria commented 1 year ago

The README is outdated with respect to the current LION version: https://github.com/HAN-Siyu/LION/blob/4ab5870550b3162174f723d485b9111ea4812ffd/README.md?plain=1#L21

amatria commented 1 year ago

Also, what does it mean that the evaluate_performance() function has been fixed? Did it produce wrong accuracies, etc.? Does it mean that the results in the paper are better?

HAN-Siyu commented 1 year ago

Also, what does it mean that the evaluate_performance() function has been fixed? Did it produce wrong accuracies, etc.? Does it mean that the results in the paper are better?

Hi @amatria,

Thanks for the interest. The original evaluate_performance() was specifically designed for the experiments in our study, and we used labels such as "Interact" and "Non-interact" to denote positive and negative classes. And evaluate_performance() always assumes the positive class is before the negative class in the confusion table. This could sometimes mistake positive class as negative class (for example, use labels 1 and 0 to denote positive and negative classes), so the metrics such as sensitivity, precision, etc will not be correct in this case. We fixed this issue in the latest version. Fortunately, this will not affect the results in the paper.

Sorry for the inconvenience, please let me know if you have further questions.

Best, Siyu

HAN-Siyu commented 1 year ago

The README is outdated with respect to the current LION version:

https://github.com/HAN-Siyu/LION/blob/4ab5870550b3162174f723d485b9111ea4812ffd/README.md?plain=1#L21

And thank you for pointing this out. Will fix it later when I'm able to access all the source data.

amatria commented 1 year ago

Hi @HAN-Siyu

I appreciate the careful response :)

Thank you very much

HAN-Siyu commented 1 year ago

Hi @HAN-Siyu

I appreciate the careful response :)

Thank you very much

Hi @amatria

Just updated links for v0.2.9. Thank you again for helping fix the issue!

HAN-Siyu commented 1 year ago

Hi @amatria,

LION has just been updated to v.0.2.9.1. The code in v0.2.9 has the same issue (mentioned above) when retraining the model of LION (using run_LION()). Please kindly use the latest version and check the LION's output if you used labels such as 0 and 1 to denote negative and positive classes. Sorry for the issue and the inaccurate results produced by LION.

Best regards, Siyu