Question about Trained Predictive Model for Major Class (Sec/SPI)

[nieuwoudtmd]

Hi USPNet Team,

I saw in the README that you have a model focused on the major class (Sec/SPI) with higher accuracy. However, I didn't see this model mentioned in your paper. Was this model used for any of the results in the paper? If not, do you have any performance results for it, especially compared to tools like SignalP?

Thanks!

Mechiel

[JunboShen]

Hi Mechiel,

Thank you for reaching out!
Yes, we do have a trained predictive model specifically targeting the major class (Sec/SPI) with higher accuracy. This model can be downloaded via the link provided in the README. However, it was not used for any of the results presented in our paper.

This model shares the same architecture as USPNet-fast and utilizes ESM-2 embeddings. To evaluate its performance, you can test it using the same code files used for the test and inference of USPNet-fast (requires organism group information). This will allow you to compare its results with other tools, such as SignalP.
If you have any further questions or need additional assistance, feel free to ask.

Thank you!