Free Software from the Zuckerman Group
Effective Sample Size Scripts: Click to Download (144 KB)
These scripts calculate the effective sample size of a given trajectory. The effective sample size is a measure of the number of statistically independent configurations in a trajectory. Please refer to the readme file in the supplied software folder for usage instructions. Faster versions available from http://loos.sourceforge.net/
Software for steady state using weighted ensemble:
Click to Download (110 KB)
Library-Based Monte Carlo Method (LBMC): Click to Download (99 MB)
This software generates an ensemble of structures given a set of atomic input coordinates from the PDB. The code swaps a predetermined number of peptide planes per step with configurations from a library of alanine, proline, and glycine. It then uses a Monte Carlo technique to accept or reject these moves. Energy is calculated using the Go method. Three other energy functions are also available for use: Ramachandran potential, Hydrogen-bonding, and Miyazawa-Jernigan contact energy. (instructions PDF)
Weighted Ensemble Approach (template):
These scripts serve as a template for the weighted ensemble approach, which is a promising path sampling method as described in http://www.pnas.org/cgi/content/abstract/104/46/18043
Decorrelation Scripts: Click to Download
These scripts implement the analysis of simulation trajectories via the effective sample size, as described in http://arxiv.org/abs/q-bio.QM/0607037. The analysis operates on Tinker style trajectory archives, and calls the Tinker subroutine ‘superpose’ for calculating RMSD’s. (go to http://dasher.wustl.edu/tinker/ to download Tinker, free of charge.) A detailed explanation of the usage and some idiosyncracies is found at the top of the file decorr-analysis.pl. Please email me with any questions or bug reports at firstname.lastname@example.org, with the subject decorrelation time analysis.
Confidence intervals for hypergeometric variance: Click version to Download –
These scripts calculate confidence intervals on estimates of the variance of a hypergeometric distribution. This allows the addition of error bars to a theoretical prediction of such a variance, so that it may be ascertained whether an observation based on finite data falls within the expected fluctuation. (explanation PDF) Please email me with any questions or bug reports at email@example.com , with the subject hyperG-CI.
Extrapolation Software: Click to Download
Nonequilibrium work values, generated for the purpose of obtaining free energy estimates with Jarzynski’s relation, provide notoriously unreliable estimates. This results from the highly nonlinear nature of the exponential average, as discussed in several of our publications. Here you can download scripts to perfrom the analysis described in a recent paper [FM Ytreberg and DM Zuckerman, J. Comp. Chem., in press, (2004)]. The scripts will provide much more reliable free energy estimates than straightforward application of Jarzynksi’s relation. They also provide auxiliary information, including finite-data block averages, to enable “manual” assessment of the estimates.