Secondary structures for validation & test sets

[zlcrrrr]

Hi,

I wonder if it is possible to add secondary structures for validation & test sets?

Thanks.

[jonathanking]

Hi Lichirui,

This is definitely possible and on my "shortlist" of things to do, but it is not the highest priority at the moment.

Part of the reason is that at evaluation time, you might assume you only know the sequence and the atomic coordinates, but I can still see how this may be useful. Another reason is that instead of using ProDy to compute secondary structure (the software package I've been using for generating/measuring data), I was able to borrow the secondary structure information shared by ProteinNet. ProteinNet only includes this data for the training set. Please stay tuned and I can update you when I have made any changes!

Best, Jonathan

[zlcrrrr]

Sounds good, thanks.

Hi Lichirui,

This is definitely possible and on my "shortlist" of things to do, but it is not the highest priority at the moment.

Part of the reason is that at evaluation time, you might assume you only know the sequence and the atomic coordinates, but I can still see how this may be useful. Another reason is that instead of using ProDy to compute secondary structure (the software package I've been using for generating/measuring data), I was able to borrow the secondary structure information shared by ProteinNet. ProteinNet only includes this data for the training set. Please stay tuned and I can update you when I have made any changes!

Best, Jonathan