OpenMM 4: A Reusable, Extensible, Hardware Independent Library for High Performance Molecular Simulation.

OpenMM is a software toolkit for performing molecular simulations on a range of high performance computing architectures. It is based on a layered architecture: the lower layers function as a reusable library that can be invoked by any application, while the upper layers form a complete environment for running molecular simulations. The library API hides all hardware-specific dependencies and optimizations from the users and developers of simulation programs: they can be run without modification on any hardware on which the API has been implemented. The current implementations of OpenMM include support for graphics processing units using the OpenCL and CUDA frameworks. In addition, OpenMM was designed to be extensible, so new hardware architectures can be accommodated and new functionality (e.g., energy terms and integrators) can be easily added.

Accelerating molecular dynamic simulation on graphics processing units.

We describe a complete implementation of all-atom protein molecular dynamics running entirely on a graphics processing unit (GPU), including all standard force field terms, integration, constraints, and implicit solvent. We discuss the design of our algorithms and important optimizations needed to fully take advantage of a GPU. We evaluate its performance, and show that it can be more than 700 times faster than a conventional implementation running on a single CPU core.

Guidelines for the routine application of the peptide hits technique.

A set of guidelines has been developed for using the peptide hits technique (PHT) as a semi-quantitative screening tool for the identification of proteins that change in abundance in a complex mixture. The dataset that formed the basis for these experiments was created using a cell lysate derived from the yeast Saccharomyces cerevisiae, spiked at various levels with serum albumin (BSA), and analyzed by LC/MS/MS and SEQUEST. Knowing that the level of only one protein (BSA) actually changed in the mixture allowed for the development and refinement of the necessary bioinformatics and statistical analyses, e.g., principal component analysis (PCA), normalization, and analysis of variation (ANOVA). As expected, the number of BSA peptide hits changed in proportion to the amount of BSA added to the sample. PCA was able to clearly distinguish between the spiked samples and the untreated sample, indicating that PCA may be able to classify samples, e.g., healthy versus diseased, in future experiments. The use of an endogenous "housekeeping" protein was found to be superior to the use of total hits for data normalization prior to analysis. An ANOVA based model readily identified BSA as a protein of interest, that is, one likely to be changing from amongst the background proteins, indicating that an ANOVA model may be able to identify individual proteins in target or biomarker discovery experiments. General guidelines based on these combined observations are set forth for future analyses and the rapid screening for candidate proteins of interest.

Changes in the protein expression of yeast as a function of carbon source.

Protein expression trends in yeast were monitored as a function of carbon source (glucose versus galactose) using multidimensional high performance liquid chromatography (HPLC) coupled to gas-phase fractionation, using relative intensity triggering (GPFri). Size exclusion HPLC was used to separate whole cell lysates, and following proteolysis of these fractions, each was separated by reversed phase HPLC, which was coupled on-line via electrospray to an ion trap mass spectrometer. The GPFri technique increased the dynamic range of proteins detected by increasing the number of peptide ions subjected to low energy collision induced dissociation to the 24 most intense ions in each of the survey scans. No protein or peptide labeling was used; instead, the number of SEQUEST identifications for each peptide (previously termed "hits") were used as a semiquantitative means of assessing both the direction (increase vs decrease) and significance of change in protein abundance. None of the traditional SEQUEST filters, e.g., Xcorr, DelCn, Sp, Rsp, etc., were employed in this study. Instead, a Student's t-test was used to distinguish those proteins that significantly and reproducibly changed between carbon sources from those that did not. This relied on the SEQUEST misassignments occurring in equal proportion between treatments and thereby negating each other; statistically significant changes in SEQUEST assignments were nonrandom events by definition and therefore reflective of correct identifications. This method of data analysis showed a large degree of concordance with results reported by other groups in similar transcriptional profiling and proteomic experiments. In all, 176 and 231 (fold-change > or = 1.1; p < or = 0.05) proteins were identified as being increased in peptide hit number when the yeast cells' source of carbon was changed between glucose and galactose, respectively.