Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Confusion with negative motifs #109

Open
PelFritz opened this issue May 31, 2023 · 4 comments
Open

Confusion with negative motifs #109

PelFritz opened this issue May 31, 2023 · 4 comments

Comments

@PelFritz
Copy link

I have a binary classifier trained and computed deepLIFT scores for just the genomic sequences that belong to class 1. My expectation was to get only positive motifs but I also get negative motifs. How does one interpret this? I consider positive motifs as those with upward facing web logos and negative motifs as those with downward facing web logos. As additional information if needed, my model ends with an output layer with a single neuron and sigmoid activation.
Any explanation to this will be appreciated.
Thanks

@akundaje
Copy link

akundaje commented May 31, 2023 via email

@PelFritz
Copy link
Author

PelFritz commented May 31, 2023

Thanks @akundaje, I think this makes sense with what we see. We could also see motifs appearing in both metaclusters (positive and negative) and suspected it was also as a result of the sequence context just like said.

Just for the sake of clarity, will it then be safe to attribute the negative motifs as those driving the decision towards the negative class (class 0) in a binary classification task?

Thanks
Fritz

@akundaje
Copy link

Yes correct. Negative motifs are favoring the negative class.

@PelFritz
Copy link
Author

Thank you @akundaje for the quick responses.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants