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
To validate the utility and quality of Cryo2StructData, we designed two deep transformer models and trained and test them on Cryo2StructData to predict backbone atoms and amino acid types from density maps.
238
238
239
239
### Deep transformer to predict protein backbone atoms and amino acid types
240
-
The inference program for the deep transformer is available in [prediction/src/infer/](prediction/src/infer/). Download the model checkpoints from [Trained on Full Cryo2StructData](https://doi.org/10.7910/DVN/SXNYRE) or [Trained on Small Subset Cryo2StructData](https://doi.org/10.7910/DVN/CGUENL) and keep them in [prediction/checkpoints](prediction/checkpoints/) directory.
240
+
The inference program for the deep transformer is available in [prediction/src/infer/](prediction/src/infer/). Download the model checkpoints from [Trained on Full Cryo2StructData](https://doi.org/10.7910/DVN/SXNYRE) or [Trained on Small Subset Cryo2StructData](https://doi.org/10.7910/DVN/DTV4JF) and keep them in [prediction/checkpoints](prediction/checkpoints/) directory.
241
241
242
242
### Hidden Markov Model (HMM) to link predicted Ca atoms into backbone structures
243
243
The Hidden Markov Model-Guided carbon-alpha atom connection program are available in [prediction/src/viterbi/](prediction/src/viterbi/). The viterbi algorithm is written in C++ program, so compile them using:
0 commit comments