Search

Scientific workflows often consist of black-box stand-alone programs that each write to one or more files, resulting in significant I/O operations. When strung together, each stage of the pipeline is bottlenecked by I/O, without much ability to modify that behavior. Since computation capabilities of large-scale distributed systems grow much faster than their I/O bandwidth, processing such big data using these workflows does not scale as the amount of data grows. While most related work focuses on data aggregation by modifying source code, we propose a Function Interception Library for Collective I/O (FILCIO), to intercept I/O function calls in C’s stdio library on a POSIX file system. This aggregates outgoing data in memory before writing to disk. Our aim is to scale I/O intensive scientific workflows without needing to access and modify their source code. In this paper, we motivate the need for I/O aggregation, and discuss what applications are poised to benefit most from FILCIO. We use an I/O intensive program where nearly 100% of its overall runtime is spent writing to disk. We show that the use of FILCIO results in a speedup of 1.75x over the native approach, and extrapolate that the best improvement is when a program generates a large volume of small writes. FILCIO is available at https://github.com/jensenq/io_aggregation

Bioinformatics software often integrates multiple off-the-shelf programs into a single compute pipeline. Each stand-alone program generates output, that is frequently saved into a plain-text file, which is then processed as input by the program that is responsible for the next stage of the computation. Modern advances in genome sequencing and protein structure resolution methods have yielded large amounts of data, that when processed by bioinformatics compute pipelines, results in vast numbers of file reads and writes. Since computation capabilities of large-scale distributed systems grow much faster than their I/O bandwidth, processing such big data using these compute pipelines will not scale as the amount of data grows. For this work, we motivate and demonstrate a dynamic interception-based I/O analysis tool to assess the file read and write characteristics of a protein mutation generation pipeline. We discuss how our analysis tool can be further extended to apply compression and in-site analysis and has the potential to scale I/O-intensive bioinformatics applications on high performance computing (HPC) systems.

Understanding how an amino acid substitution affects a protein's stability can aid in the design of pharmaceutical drugs that aim to counter the deleterious effects caused by protein mutants. Unfortunately, performing mutation experiments on the physical protein is both time and cost prohibitive. Thus an exhaustive analysis which includes systematically mutating all amino acids in the physical protein is infeasible. Computational methods have been developed over the years to predict the effects of mutations, but even many of them are computationally intensive else are dependent on homology or experimental data that may not be available for the protein being studied. In this work we motivate and present a computation pipeline whose only input is a Protein Data Bank file containing the 3D coordinates of the atoms of a biomolecule. Our high-throughput approach uses our rMutant algorithm to exhaustively generate in silico mutants with amino acid substitutions to Glycine, Alanine, and Serine for all residues in a protein. We exploit the speed of a fast rigidity analysis approach to analyze our protein variants, and develop a Mutation Sensitivity (MuSe) Map to identify residues that are most sensitive to mutations. We present three case studies and show the degree to which a MuSe Map is able to identify those amino acids which are susceptible to the effects of mutations.

Predicting how amino acid substitutions affect the stability of a protein has relevance to drug design and may help elucidate the mechanisms of disease-causing protein variants. Unfortunately, wet-lab experiments are time intensive, and to the best of our knowledge there are no efficient computational techniques to asses the effect of multiple mutations. In this work we present a new approach for inferring the effects of single and multiple mutations on a protein's structure. Our rMutant algorithm generates in silico mutants with single or multiple amino acid substitutions. We use a graph-theoretic rigidity analysis approach to compute the distributions of rigid cluster sizes of the wild type and mutant structures which we then analyze to infer the effect of the amino acid substitutions. We successfully predict the effects of multiple mutations for which our previous methods were unsuccessful. We validate the predictions of our computational approach against experimental ΔΔG data. To demonstrate the utility of using rigid cluster size distributions to infer the effects of mutations, we also present a Random Forest Machine Learning approach that relies on rigidity data to predict which residues are critical to the stability of a protein. We predict the destabilizing effects of a single or multiple mutations with over 86% accuracy.

We initiate in silico rigidity-theoretical studies of protein crystal structures, with the goal to determine if, and how, the interactions among neighboring crystal cells affect the flexibility of biological unit. We use an efficient graph-based algorithm for rigidity analysis, and other tools available through the KINARI-Web server developed in our group. For the RNase A protein (PDB file 5RSA), which has the remarkable property of being functionally active even when crystallized, we found that the individual protein and its crystal form retain the flexibility parameters between the two states. By contrast, other proteins in our data set aggregated in larger rigid clusters when analyzed as crystals.

Predicting the effect of a single amino acid substitution on the stability of a protein structure is a fundamental task in macromolecular modeling. It has relevance to drug design and understanding of disease-causing protein variants. We present KINARI-Mutagen, a web server for performing in silico mutation experiments on protein structures from the Protein Data Bank. Our rigidity-theoretical approach permits fast evaluation of the effects of mutations that may not be easy to perform in vitro, because it is not always possible to express a protein with a specific amino acid substitution. We use KINARI-Mutagen to identify critical residues, and we show that our predictions correlate with destabilizing mutations to glycine. In two in-depth case studies we show that the mutated residues identified by KINARI-Mutagen as critical correlate with experimental data, and would not have been identified by other methods such as Solvent Accessible Surface Area measurements or residue ranking by contributions to stabilizing interactions. We also generate 48 mutants for 14 proteins, and compare our rigidity-based results against experimental mutation stability data. KINARI-Mutagen is available at http://kinari.cs.umass.edu .

Proteins are the fundamental building blocks of all biological systems. To perform their function, proteins generally undergo self-motions that result in changes in their three-dimensional shape. In order to understand the function of a protein and thus to be able to infer how to therapeutically regulate its function, it is necessary to have detailed knowledge of the feasible self-motions of the protein. Such knowledge cannot be obtained by existing experimental methods. In this paper, we present preliminary evidence that accurate and computationally efficient simulation of the self-motions of a protein may be achieved by partitioning the simulation based on the type of self-motions. In support of this view, we present a method and accompanying simulation results that the large-scale motions of a protein can be simulated based entirely on kinematic considerations. The proposed method leverages insights from kinematics and operational space control from robotics. We believe the proposed method to be a first step towards a general, accurate, and efficient method for the simulation of protein motion.