- JEDi is a powerful tool for inspecting the dynamics of proteins from trajectories derived from MD or Geometric simulations, or NMA.
- All JEDi features are controlled via an input file containing a list of KEY=VALUE pairs
- There are 5 Levels of Coarse-Graining:
- All-Atom by residue
- Heavy-Atom by residue
- Back-Bone by residue
- Alpha-Carbon by residue
- Atom List
- There are 4 Types of PCA:
- Hierarchical Cartesian
- Direct Cartesian
- Distance-Pair
- Residue Pair Interaction
- There are 3 PCA Models:
- Covariance
- Correlation
- Partial-Correlation
- There is the option to perform Outlier Processing:
- Outliers Removal is perfomed at a specified cutoff
- Outliers Selection is perfomed at a specified cutoff
- The results of outlier removal and selection are compared
- There is the option to perform Sparsification of the Correlation and Partial Correlation Matrices:
- Separate cutoffs are provided for each matrix
- When both Correlation and Partial Correlation models are selected, an adjacency matrix is provided to help identify activator and suppressor variables
- The results of sparse and non-sparse analyses are compared
- All JEDi output is provided as high quality PNG images for immediate inspection:
- PCs and DVPs
- Square Modes
- Scree Plots
- Variable statistics
- MSA scores
- Eigenvector Collectivity
- Iterated RMSIPs with comparison to random and with Z-scores for significance
- Kernel PCA post-processing of the reduced data is available, using an assortment of kernels including:
- Mutual Information
- Neural-Net (Tanh)
- Gaussian
- Log
- Circular
- Mahalanobis
- Euclidean
- Polynomial Degrees 2, 3, 4, and cross XY
- Linear
- JEDi uses subspace analysis to assess the similarity of essential vector spaces defined by each dynamical method:
- Inter-Model within each type of PCA
- Inter-Type when subsets yield subspaces with matching dimensions
- JEDi computes the Free Energy of a trajectory from two PCs (normed projections) so that a FE surface can be constructed.
- JEDi produces PDB frames that capture the PCA modes and script files for viewing those frames in PyMol(TM) as movies:
- Individual modes
- The superposition of the Top Essential Modes
- For each JEDi run, a comprehensive log file is generated with time stamp recoding the parameters and key results of each analysis
-
Please report bugs by submitting a new issue tagged with BUG:
- Be sure to provide enough data to reproduce the error
- Include JED/JEDi input files, JED/JEDi errors, and associated Java stack traces
-
Please suggest new features by submitting a new issue tagged with NEW FEATURE or ENHANCEMENT
-
Please contact me if you wish to become a collaborator on this software:
-
Check out this link for a demonstration video for using JEDi:
Charles David: charles.david@plantandfood.co.nz
- New Java Docs
- New executable JAR files