Asset Detail:
Autoencoder for MD Simulation Data (P2B1)
Asset Detail:
Autoencoder for MD Simulation Data (P2B1)
Overview
ASSET LINK: | https://modac.cancer.gov/assetDetails?dme_data_id=NCI-DME-MS01-7681692 |
PROGRAM NAME: | NCI-DOE Collaboration |
STUDY NAME: | NCI-DOE-Collaboration ADMIRRAL project: Improving Outcomes for RAS-related Cancers |
ASSET NAME: | Autoencoder for MD Simulation Data (P2B1) |
ASSET PATH: | /NCI_DOE_Archive/JDACS4C/JDACS4C_Pilot_2/autoencoder_md_simulation_data_p2b1 |
Asset Attributes
ATTRIBUTE | VALUE |
---|---|
ASSET NAME | Autoencoder for MD Simulation Data (P2B1) |
ASSET DESCRIPTION | This collection contains the model topology files and the trained model weights for the Autoencoder for MD Simulation Data (P2B1). P2B1 is an autoencoder for molecular dynamics (MD) simulation data that compresses the output of MD runs into a minimal set of features that can be used as tractable input into downstream algorithms. This collection includes the full autoencoder (p2b1.model.json), just the encoder part of the autoencoder (p2b1.encoder.json), and the corresponding trained model weights after two epochs (model_weights.hdf5 and encoder_weights.hdf5, respectively) as sample weights corresponding to the simulation input data. The autoencoder is a convolutional neural network. The authors of this model trained it on the MD output of a 3k disordered, 3-component-system (DPPC-DOPC-CHOL). They ran this model for two epochs by default, producing a converged reconstruction loss of ~0.0314. For more information, refer to the Simulation Input Data link and the GitHub Repository link. |
ASSET IDENTIFIER | autoencoder_md_simulation_data_p2b1 |
ASSET TYPE | Model |
DOI | NA |
MODEL DOMAIN | Autoencoder |
MODEL FRAMEWORK | Keras |
MODEL PLATFORM | None |
POC NAME | Andrew Weisman |
POC EMAIL | andrew.weisman@nih.gov |
IS MODEL DEPLOYED | No |
COLLECTION SIZE | 101.5 MB |
GITHUB REPOSITORY | https://github.com/CBIIT/NCI-DOE-Collab-Pilot2-Autoencoder_MD_Simulation_Data |
SIMULATION INPUT DATA | https://modac.cancer.gov/searchTab?dme_data_id=NCI-DME-MS01-7654212 |
Asset Files
To download files, please login.
FILE/COLLECTION | FILE SIZE | ACTIONS |
---|