Christopher James Langmead

A Bayesian Approach to Model Checking Biological Systems

Recently, there has been considerable interest in the use of Model Checking for Systems Biology. Unfortunately, the state space of stochastic biological models is often too large for classical Model Checking techniques. For these models, a statistical approach to Model Checking has been shown to be an effective alternative. Extending our earlier work, we present the first algorithm for performing statistical Model Checking using Bayesian Sequential Hypothesis Testing. We show that our Bayesian approach outperforms current statistical Model Checking techniques, which rely on tests from Classical (aka Frequentist) statistics, by requiring fewer system simulations. Another advantage of our approach is the ability to incorporate prior Biological knowledge about the model being verified. We demonstrate our algorithm on a variety of models from the Systems Biology literature and show that it enables faster verification than state-of-the-art techniques, even when no prior knowledge is available.

Comparison of Existing Clinical Scoring Systems to Predict Persistent Organ Failure in Patients With Acute Pancreatitis

BACKGROUND & AIMS It is important to identify patients with acute pancreatitis who are at risk for developing persistent organ failure early in the course of disease. Several scoring systems have been developed to predict which patients are most likely to develop persistent organ failure. We head-to-head compared the accuracy of these systems in predicting persistent organ failure, developed rules that combined these scores to optimize predictive accuracy, and validated our findings in an independent cohort. METHODS Clinical data from 2 prospective cohorts were used for training (n = 256) and validation (n = 397). Persistent organ failure was defined as cardiovascular, pulmonary, and/or renal failure that lasted for 48 hours or more. Nine clinical scores were calculated when patients were admitted and 48 hours later. We developed 12 predictive rules that combined these scores, in order of increasing complexity. RESULTS Existing scoring systems showed modest accuracy (areas under the curve at admission of 0.62-0.84 in the training cohort and 0.57-0.74 in the validation cohort). The Glasgow score was the best classifier at admission in both cohorts. Serum levels of creatinine and blood urea nitrogen provided similar levels of discrimination in each set of patients. Our 12 predictive rules increased accuracy to 0.92 in the training cohort and 0.84 in the validation cohort. CONCLUSIONS The existing scoring systems seem to have reached their maximal efficacy in predicting persistent organ failure in acute pancreatitis. Sophisticated combinations of predictive rules are more accurate but cumbersome to use, and therefore of limited clinical use. Our ability to predict the severity of acute pancreatitis cannot be expected to improve unless we develop new approaches.

Serum Biomarker Panels for the Detection of Pancreatic Cancer

Purpose: Serum–biomarker based screening for pancreatic cancer could greatly improve survival in appropriately targeted high-risk populations. Experimental Design: Eighty-three circulating proteins were analyzed in sera of patients diagnosed with pancreatic ductal adenocarcinoma (PDAC, n = 333), benign pancreatic conditions (n = 144), and healthy control individuals (n = 227). Samples from each group were split randomly into training and blinded validation sets prior to analysis. A Metropolis algorithm with Monte Carlo simulation (MMC) was used to identify discriminatory biomarker panels in the training set. Identified panels were evaluated in the validation set and in patients diagnosed with colon (n = 33), lung (n = 62), and breast (n = 108) cancers. Results: Several robust profiles of protein alterations were present in sera of PDAC patients compared to the Healthy and Benign groups. In the training set (n = 160 PDAC, 74 Benign, 107 Healthy), the panel of CA 19–9, ICAM-1, and OPG discriminated PDAC patients from Healthy controls with a sensitivity/specificity (SN/SP) of 88/90%, while the panel of CA 19–9, CEA, and TIMP-1 discriminated PDAC patients from Benign subjects with an SN/SP of 76/90%. In an independent validation set (n = 173 PDAC, 70 Benign, 120 Healthy), the panel of CA 19–9, ICAM-1 and OPG demonstrated an SN/SP of 78/94% while the panel of CA19–9, CEA, and TIMP-1 demonstrated an SN/SP of 71/89%. The CA19–9, ICAM-1, OPG panel is selective for PDAC and does not recognize breast (SP = 100%), lung (SP = 97%), or colon (SP = 97%) cancer. Conclusions: The PDAC-specific biomarker panels identified in this investigation warrant additional clinical validation to determine their role in screening targeted high-risk populations. Clin Cancer Res; 17(4); 805–16. ©2010 AACR.

Learning generative models for protein fold families

We introduce a new approach to learning statistical models from multiple sequence alignments (MSA) of proteins. Our method, called GREMLIN (Generative REgularized ModeLs of proteINs), learns an undirected probabilistic graphical model of the amino acid composition within the MSA. The resulting model encodes both the position‐specific conservation statistics and the correlated mutation statistics between sequential and long‐range pairs of residues. Existing techniques for learning graphical models from MSA either make strong, and often inappropriate assumptions about the conditional independencies within the MSA (e.g., Hidden Markov Models), or else use suboptimal algorithms to learn the parameters of the model. In contrast, GREMLIN makes no a priori assumptions about the conditional independencies within the MSA. We formulate and solve a convex optimization problem, thus guaranteeing that we find a globally optimal model at convergence. The resulting model is also generative, allowing for the design of new protein sequences that have the same statistical properties as those in the MSA. We perform a detailed analysis of covariation statistics on the extensively studied WW and PDZ domains and show that our method out‐performs an existing algorithm for learning undirected probabilistic graphical models from MSA. We then apply our approach to 71 additional families from the PFAM database and demonstrate that the resulting models significantly out‐perform Hidden Markov Models in terms of predictive accuracy. Proteins 2011;

Statistical Model Checking in BioLab: Applications to the Automated Analysis of T-Cell Receptor Signaling Pathway

We present an algorithm, called BioLab , for verifying temporal properties of rule-based models of cellular signalling networks. BioLab models are encoded in the BioNetGen language, and properties are expressed as formulae in probabilistic bounded linear temporal logic. Temporal logic is a formalism for representing and reasoning about propositions qualified in terms of time. Properties are then verified using sequential hypothesis testing on executions generated using stochastic simulation. BioLab is optimal, in the sense that it generates the minimum number of executions necessary to verify the given property. BioLab also provides guarantees on the probability of it generating Type-I (i.e., false-positive) and Type-II (i.e., false-negative) errors. Moreover, these error bounds are pre-specified by the user. We demonstrate BioLab by verifying stochastic effects and bistability in the dynamics of the T-cell receptor signaling network.

A polynomial-time nuclear vector replacement algorithm for automated NMR resonance assignments.

High-throughput NMR structural biology can play an important role in structural genomics. We report an automated procedure for high-throughput NMR resonance assignment for a protein of known structure, or of a homologous structure. These assignments are a prerequisite for probing protein-protein interactions, protein-ligand binding, and dynamics by NMR. Assignments are also the starting point for structure determination and refinement. A new algorithm, called Nuclear Vector Replacement (NVR) is introduced to compute assignments that optimally correlate experimentally measured NH residual dipolar couplings (RDCs) to a given a priori whole-protein 3D structural model. The algorithm requires only uniform( 15)N-labeling of the protein and processes unassigned H(N)-(15)N HSQC spectra, H(N)-(15)N RDCs, and sparse H(N)-H(N) NOEs (d(NN)s), all of which can be acquired in a fraction of the time needed to record the traditional suite of experiments used to perform resonance assignments. NVR runs in minutes and efficiently assigns the (H(N),(15)N) backbone resonances as well as the d(NN)s of the 3D (15)N-NOESY spectrum, in O(n(3)) time. The algorithm is demonstrated on NMR data from a 76-residue protein, human ubiquitin, matched to four structures, including one mutant (homolog), determined either by x-ray crystallography or by different NMR experiments (without RDCs). NVR achieves an assignment accuracy of 92-100%. We further demonstrate the feasibility of our algorithm for different and larger proteins, using NMR data for hen lysozyme (129 residues, 97-100% accuracy) and streptococcal protein G (56 residues, 100% accuracy), matched to a variety of 3D structural models. Finally, we extend NVR to a second application, 3D structural homology detection, and demonstrate that NVR is able to identify structural homologies between proteins with remote amino acid sequences using a database of structural models.

Symbolic Approaches for Finding Control Strategies in Boolean Networks

We present an exact algorithm, based on techniques from the field of Model Checking, for finding control policies for Boolean Networks (BN) with control nodes. Given a BN, a set of starting states, I, a set of goal states, F, and a target time, t, our algorithm automatically finds a sequence of control signals that deterministically drives the BN from I to F at, or before time t, or else guarantees that no such policy exists. Despite recent hardness-results for finding control policies for BNs, we show that, in practice, our algorithm runs in seconds to minutes on over 13,400 BNs of varying sizes and topologies, including a BN model of embryogenesis in Drosophila melanogaster with 15,360 Boolean variables. We then extend our method to automatically identify a set of Boolean transfer functions that reproduce the qualitative behavior of gene regulatory networks. Specifically, we automatically learn a BN model of D. melanogaster embryogenesis in 5.3 seconds, from a space containing 6.9 x 10(10) possible models.

Angiopoietin-2, a Regulator of Vascular Permeability in Inflammation, Is Associated With Persistent Organ Failure in Patients With Acute Pancreatitis From the United States and Germany

OBJECTIVES:Patients with severe acute pancreatitis (AP) typically develop vascular leak syndrome, resulting in hemoconcentration, hypotension, pulmonary edema, and renal insufficiency. Angiopoietin-1 (Ang-1) and 2 (Ang-2) are autocrine peptides that reduce or increase endothelial permeability, respectively. The aim of this study was to determine whether Ang-1 and/or Ang-2 levels are predictive biomarkers of persistent organ failure (>48 h) and prolonged hospital course.METHODS:Banked serum from 28 patients enrolled in the Severity of Acute Pancreatitis Study at the University of Pittsburgh Medical Center (UPMC) and 58 controls was analyzed for Ang-1 and Ang-2 levels. Separately, serum from 123 patients and 103 controls at Greifswald University (GU), Germany was analyzed for Ang-2 levels. Angiopoietin levels were measured by enzyme-linked immunosorbent assay.RESULTS:In all, 6 out of 28 UPMC patients (21%) and 14 out of 123 GU patients (13%) developed persistent organ failure and were classified as severe AP. Ang-2 was significantly higher on admission in patients who developed persistent organ failure compared with those who did not in UPMC (3,698 pg/ml vs. 1,001 pg/ml; P=0.001) and GU (4,945 pg/ml vs. 2,631 pg/ml; P=0.0004) cohorts. After data scaling, admission Ang-2 levels showed a receiver-operator curve of 0.81, sensitivity 90%, and specificity 67% in predicting persistent organ failure. In addition, Ang-2 levels remained significantly higher in severe AP compared with mild AP patients until day 7 (days 2–4: P<0.005; day 7: P<0.02). Ang-1 levels were not significantly different between mild and severe AP patients on admission.CONCLUSIONS:Elevated serum Ang-2 levels on admission are associated with and may be a useful biomarker of predicting persistent organ failure and ongoing endothelial cell activation in AP.

Dynamic allostery governs cyclophilin A-HIV capsid interplay

Significance The mechanisms of how Cyclophilin A (CypA) regulates HIV-1 infectivity remain poorly understood. We examined the role of dynamics in capsid (CA) protein assemblies by magic-angle-spinning NMR. The assembled CA is highly dynamic. Dipolar tensors calculated from molecular dynamics trajectories are in quantitative agreement with the NMR results. Motions in the CypA loop are sequence-dependent and attenuated in the escape mutants A92E and G94D. Dynamics are similar in escape mutants and CA/CypA complex. These findings suggest that CA escapes from CypA dependence through dynamic allostery. Thus, a host factors function in HIV infectivity may not be primarily associated with a structural change of the capsid core, but with altering its dynamics, such as the reduction of motions for the CypA loop. Host factor protein Cyclophilin A (CypA) regulates HIV-1 viral infectivity through direct interactions with the viral capsid, by an unknown mechanism. CypA can either promote or inhibit viral infection, depending on host cell type and HIV-1 capsid (CA) protein sequence. We have examined the role of conformational dynamics on the nanosecond to millisecond timescale in HIV-1 CA assemblies in the escape from CypA dependence, by magic-angle spinning (MAS) NMR and molecular dynamics (MD). Through the analysis of backbone 1H-15N and 1H-13C dipolar tensors and peak intensities from 3D MAS NMR spectra of wild-type and the A92E and G94D CypA escape mutants, we demonstrate that assembled CA is dynamic, particularly in loop regions. The CypA loop in assembled wild-type CA from two strains exhibits unprecedented mobility on the nanosecond to microsecond timescales, and the experimental NMR dipolar order parameters are in quantitative agreement with those calculated from MD trajectories. Remarkably, the CypA loop dynamics of wild-type CA HXB2 assembly is significantly attenuated upon CypA binding, and the dynamics profiles of the A92E and G94D CypA escape mutants closely resemble that of wild-type CA assembly in complex with CypA. These results suggest that CypA loop dynamics is a determining factor in HIV-1s escape from CypA dependence.

Free energy estimates of all-atom protein structures using generalized belief propagation

We present a technique for approximating the free energy of protein structures using Generalized Belief Propagation (GBP). The accuracy and utility of these estimates are then demonstrated in two different application domains. First, we show that the entropy component of our free energy estimates can be useful in distinguishing native protein structures from decoys -- structures with similar internal energy to that of the native structure, but otherwise incorrect. Our method is able to correctly identify the native fold from among a set of decoys with 87.5% accuracy over a total of 48 different immunoglobin folds. The remaining 12.5% of native structures are ranked among the top 4 of all structures. Second, we show that our estimates of ΔΔG upon mutation upon mutation for three different data sets have linear correlations between 0.63-0.70 with experimental values and statistically significant p-values. Together, these results suggests that GBP is an effective means for computing free energy in all-atom models of protein structures. GBP is also efficient, taking a few minutes to run on a typical sized protein, further suggesting that GBP may be an attractive alternative to more costly molecular dynamic simulations for some tasks.