Clustering & PCA
Source: src/lib/md/MdClusteringPanel.svelte
Overview
Clusters MD trajectory frames based on structural similarity and performs PCA for dimensionality reduction. Identifies distinct conformational states.
Components
MdClusteringPanel
Interactive panel for clustering and PCA configuration.
Algorithms
K-Means
Partition frames into k clusters based on structural descriptors.
DBSCAN
Density-based clustering that automatically determines the number of clusters.
Hierarchical
Agglomerative clustering with dendrogram visualization.
PCA
Principal Component Analysis
Projects high-dimensional trajectory data onto orthogonal components capturing maximum variance.
Visualization
2D scatter plot of frames in PC1-PC2 space, colored by cluster assignment.
Server API
Endpoint: POST /api/md/clustering