@@ -42,7 +42,7 @@ The installation procedure works correctly with recent versions of Linux (Ubuntu
## DEMO (data example)
To test WARIO, we provide a [MD simulation](https://gitlab.laas.fr/moma/methods/analysis/WARIO/-/tree/main/wario/P113) of the peptide P113 using the CHARMM36IDPFF force-field. Data have been extracted from [Jephthah _et al_ (2021)](https://doi.org/10.1021/acs.jctc.1c00408), where details on the simulation and the force-field can be found. The P113 ensemble contains $n=100001$ conformations for a sequence of length $L=12$. To run WARIO on this data, the user can uncomment the lines
To test WARIO, we provide a dataset corresponding to the conformational ensemble of a small peptide, P-113, obtained from [MD simulations](https://gitlab.laas.fr/moma/methods/analysis/WARIO/-/tree/main/wario/P113). Data was extracted from [(Jephthah _et al_, 2021)](https://doi.org/10.1021/acs.jctc.1c00408), where details on the simulation and the force-field can be found. The P-113 ensemble contains $n=100001$ conformations for a sequence of length $L=12$. To run WARIO on this data, the user can uncomment the lines
```
#ensemble_folder = "/".join([path_to_notebook,'P113']) # Path to the folder containing trajectory files
...
...
@@ -50,7 +50,7 @@ To test WARIO, we provide a [MD simulation](https://gitlab.laas.fr/moma/methods/
```
in the second code cell of the [contact_clustering](https://gitlab.laas.fr/moma/methods/analysis/WARIO/-/blob/main/wario/contact_clustering.ipynb) notebook, and proceed following the usual guidelines. Then, results will be automatically saved in the [P113 folder](https://gitlab.laas.fr/moma/methods/analysis/WARIO/-/tree/main/wario/P113). The typical running time for this data example is around 30 minutes in a normal desktop computer using 5 threads.
By setting the clustering precision at the default level (i.e. minimum cluster size = 1% of the total number of conformations), WARIO partitions the P113 ensemble into 12 clusters and classifies the 93.04% of conformations. The ensemble is characterized by the following family of weighted cluster-specific contact maps:
By setting the clustering precision at the default level (i.e. minimum cluster size = 1% of the total number of conformations), WARIO partitions the P-113 ensemble into 12 clusters and classifies the 93.04% of conformations. The ensemble is characterized by the following family of weighted cluster-specific contact maps:
<br/>
<br/>
...
...
@@ -73,8 +73,8 @@ By setting the clustering precision at the default level (i.e. minimum cluster s
<br/>
<br/>
Each cluster can be complementary described by the average DSSP propensities of its conformations. The corresponding plots for P113 are available [here](https://gitlab.laas.fr/moma/methods/analysis/WARIO/-/tree/main/wario/demo).
Structural properties such as the average DSSP propensities of each cluster can be then computed. The corresponding plots for P-113 are available [here](https://gitlab.laas.fr/moma/methods/analysis/WARIO/-/tree/main/wario/demo).
## Large ensembles
The current implementation of WARIO can be run in a normal desktop computer for ensembles with $nL\lessapprox 3\cdot 10^7$. Limitations are mainly due to memory constraints. We are currently working on a more efficient implementation for large ensembles, that can be run in remote servers for larger ensembles. Until it is released, we can [provide guidance](mailto:javier.gonzalezdelgado@mcgill.ca) on how to run the current code in limiting settings $nL\approx 3-4\cdot 10^7$ more efficiently.
The current implementation of WARIO can be run in a normal desktop computer for ensembles with $nL\lessapprox 3\cdot 10^7$. Limitations are mainly due to memory constraints. We are currently working on a more efficient implementation for large ensembles, which can be run in remote servers. Until it will be released, we can [provide guidance](mailto:javier.gonzalezdelgado@mcgill.ca) on how to run the current code in limiting settings $nL\approx 3-4\cdot 10^7$ more efficiently.
