NetSci: A Library\nfor High Performance Biomolecular\nSimulation Network Analysis Computation

We present the NetSci program–an open-source scientific\nsoftware package designed for estimating mutual information (MI) between\ndata sets using GPU acceleration and a k-nearest-neighbor algorithm.\nThis approach significantly enhances calculation speed, achieving\nimprovements of several orders of magnitude over traditional CPU-based\nmethods, with data set size limits dictated only by available hardware.\nTo validate NetSci, we accurately compute MI for an analytically verifiable\ntwo-dimensional Gaussian distribution and replicate the generalized\ncorrelation (GC) analysis previously conducted on the B1 domain of\nprotein G. We also apply NetSci to molecular dynamics simulations\nof the Sarcoendoplasmic Reticulum Calcium-ATPase (SERCA) pump, exploring\nthe allosteric mechanisms and pathways influenced by ATP and 2′-deoxy-ATP\n(dATP) binding. Our analysis reveals distinct allosteric effects induced\nby ATP compared to dATP, with predicted information pathways from\nthe bound nucleotide to the calcium-binding domain differing based\non the nucleotide involved. NetSci proves to be a valuable tool for\nestimating MI and GC in various data sets and is particularly effective\nfor analyzing intraprotein communication and information transfer.