Biology subjects

Raman, S.

Publications and source records attributed to Raman, S..

44 records · Page 3Linked to original sources

bioRxiv2020

An Introduction to Thermodynamic Integration and Application to Dynamic Causal Models

In generative modeling of neuroimaging data, such as dynamic causal modeling (DCM), one typically considers several alternative models, either to determine the most plausible explanation for observed data (Bayesian model selection) or to account for model uncertainty (Bayesian model averaging). Both procedures rest on estimates of the model evidence, a principled trade-off between model accuracy and complexity. In the context of DCM, the log evidence is usually approximated using variational Bayes. Although this approach is highly efficient, it makes distributional assumptions and is vulnerable to local extrema. This paper introduces the use of thermodynamic integration (TI) for Bayesian model selection and averaging in the context of DCM. TI is based on Markov chain Monte Carlo sampling which is asymptotically exact but orders of magnitude slower than variational Bayes. In this paper, we explain the theoretical foundations of TI, covering key concepts such as the free energy and its origins in statistical physics. Our aim is to convey an in-depth understanding of the method starting from its historical origin in statistical physics. In addition, we demonstrate the practical application of TI via a series of examples which serve to guide the user in applying this method. Furthermore, these examples demonstrate that, given an efficient implementation and hardware capable of parallel processing, the challenge of high computational demand can be overcome successfully. The TI implementation presented in this paper is freely available as part of the open source software TAPAS. Author summaryWhen fitting computational models to data in the setting of Bayesian inference, a user has the choice between two broad classes of algorithms: variational inference and Monte Carlo simulation. While both methods have advantages and drawbacks, variational inference has become standard in the domain of modelling directed brain connectivity due to its computational efficiency, especially when the challenges to select between competing hypotheses that explain the observed data. By contrast, the high computational demand by Monte Carlo methods has so far prevented their widespread use for inference on brain connectivity, despite their capability to overcome some of the shortcomings of variational inference. In this paper, we introduce the user to thermodynamic integration (TI), a Monte Carlo method designed for model fitting and model selection. By covering its foundations and historical origins in statistical physics, we hope to convey an in-depth understanding of TI that goes beyond a purely technical treatment. In addition, we also provide examples for concrete applications, demonstrating that, given an efficient implementation and up-to-date hardware, the challenge of high computational demand can be overcome successfully.

neuroscience↗

bioRxiv2020

Epistasis shapes the fitness landscape of an allosteric specificity switch

Epistasis is a major determinant in the emergence of novel protein function. In allosteric proteins, direct interactions between inducer-binding mutations propagate through the allosteric network, manifesting as epistasis at the level of biological function. Elucidating this relationship between local interactions and their global effects is essential to understanding evolution of allosteric proteins. We integrate computational design, structural and biophysical analysis to characterize the emergence of novel inducer specificity in an allosteric transcription factor. Adaptive landscapes of different inducers of the engineered mutant show that a few strong epistatic interactions constrain the number of viable sequence pathways, revealing ridges in the fitness landscape leading to new specificity. Crystallographic evidence shows a single mutation drives specificity by reshaping the binding pocket. Comparison of biophysical and functional landscapes emphasizes the nonlinear relationship between local inducer affinity and global function (allostery). Our results highlight the functional and evolutionary complexity of allosteric proteins.

biochemistry↗

bioRxiv2020

Mapping the Functional Landscape of the Receptor Binding Domain of T7 Bacteriophage by Deep Mutational Scanning

The interaction between a bacteriophage and its host is mediated by the phages receptor binding protein (RBP). Despite its fundamental role in governing phage activity and host range, the molecular rules of RBP function remain a mystery. Here, we systematically dissect the functional role of every residue in the tip domain of T7 phage RBP using a novel phage genome engineering method called ORACLE (Optimized Recombination, Accumulation and Library Expression). ORACLE is a high-throughput, locus-specific, sequence-programmable method to create a large, unbiased library of phage variants at a targeted gene locus. Using ORACLE, we generated all single amino acid substitutions at every site (1660 variants) of the tip domain to quantify the functional role of all variants on multiple bacterial hosts. This rich dataset allowed us to cross compare functional profiles of each host to precisely identify regions of functional importance, many which were previously unknown. Host-specific substitution patterns displayed differences in site specificity and physicochemical properties of mutations indicating exquisite adaptation to individual hosts. Comparison of enriched variants across hosts also revealed a tradeoff between activity and host range. We discovered gain-of-function variants effective against resistant hosts and host-constricting variants that selectively eliminated certain hosts. We demonstrate therapeutic utility against uropathogenic E. coli by engineering a highly active T7 variants to avert emergence of spontaneous resistance of the pathogen. Our approach presents a generalized framework for systematic and comprehensive characterization of sequence-function relationships in phages on an unprecedented scale.

synthetic biology↗

bioRxiv2020

Heterologous production of 1-tuberculosinyladenosine in Mycobacterium kansasii models pathoevolution towards the transcellular lifestyle of Mycobacterium tuberculosis.

Mycobacterium kansasii is an environmental non-tuberculous mycobacterium that causes opportunistic tuberculosis-like disease. It is one of the most closely related species to the M. tuberculosis complex. Using M. kansasii as a proxy for the M. kansasii-M. tuberculosis-common ancestor, we asked whether introducing the M. tuberculosis-specific gene pair Rv3377c-Rv3378c into M. kansasii affects the course of experimental infection. Expression of these genes resulted in the production of an adenosine-linked lipid species, known as 1-tuberculosinyladenosine (1-TbAd), but did not alter growth in vitro under standard conditions. Production of 1-TbAd enhanced growth of M. kansasii under acidic conditions through a bacterial cell-intrinsic mechanism independent of controlling pH in the bulk extracellular and intracellular spaces. Production of 1-TbAd led to greater burden of M. kansasii in the lung of C57Bl/6 mice during the first 24 hours after infection and ex vivo infections of alveolar macrophages recapitulated this phenotype within the same time frame. However, in long-term infections, production of 1-TbAd resulted in impaired bacterial survival in both C57Bl/6 mice and Ccr2-/- mice. We have demonstrated that M. kansasii is a valid surrogate of M. tuberculosis to study virulence factors acquired by the latter organism, yet shown the challenge inherent to studying the complex evolution of mycobacterial pathogenicity with isolated gene complementation. IMPORTANCEThis work sheds light on the role of the lipid 1-tuberculosinyladenosine in the evolution of an environmental ancestor to M. tuberculosis. On a larger scale, it reinforces the importance of horizontal gene transfer in bacterial evolution and examines novel models and methods to provide a better understanding of the subtle effects of individual M. tuberculosis-specific virulence factors in infection settings that are relevant to the pathogen.

microbiology↗

bioRxiv2020

Functional Plasticity and Evolutionary Adaptation of Allosteric Regulation

Allostery is a fundamental regulatory mechanism of protein function. Despite notable advances, understanding the molecular determinants of allostery remains an elusive goal. Our current knowledge of allostery is principally shaped by a structure-centric view which makes it difficult to understand the decentralized character of allostery. We present a function-centric approach using deep mutational scanning to elucidate the molecular basis and underlying functional landscape of allostery. We show that allosteric signaling exhibits a high-degree of functional plasticity and redundancy through myriad mutational pathways. Residues critical for allosteric signaling are surprisingly poorly conserved while those required for structural integrity are highly conserved, suggesting evolutionary pressure to preserve fold over function. Our results suggest multiple solutions to the thermodynamic conditions of cooperativity, in contrast to the common view of a finely-tuned allosteric residue network maintained under selection.

biochemistry↗

bioRxiv2019

Identification of the C. sordellii lethal toxin receptor elucidates principles of receptor specificity in clostridial toxins

Clostridium sordellii lethal toxin (TcsL) is responsible for an almost invariably lethal toxic shock syndrome associated with gynecological C. sordellii infections. Here, using CRISPR/Cas9 screening, we identify semaphorins SEMA6A and SEMA6B as the cellular receptors for TcsL and demonstrate that soluble extracellular SEMA6A can protect mice from TcsL-induced edema. A 3.3 [A] cryo-EM structure shows that TcsL binds SEMA6A with the same region that the highly related C. difficile TcdB toxin uses to bind structurally unrelated Frizzled receptors. Remarkably, reciprocal mutations in this evolutionarily divergent surface are sufficient to switch receptor specificity between the toxins. Our findings establish semaphorins as physiologically relevant receptors for TcsL, and reveal the molecular basis for the difference in tissue targeting and disease pathogenesis between highly related toxins.

molecular biology↗

bioRxiv2019

Design of a transcriptional biosensor for the portable, on-demand detection of cyanuric acid

Rapid molecular biosensing is an emerging application area for synthetic biology. Here, we engineer a portable biosensor for cyanuric acid (CYA), an analyte of interest for human and environmental health, using a LysR-type transcription regulator (LTTR) from Pseudomonas within the context of Escherichia coli gene expression machinery. To overcome cross-host portability challenges of LTTRs, we rationally engineered hybrid Pseudomonas-E. coli promoters by integrating DNA elements required for transcriptional activity and ligand-dependent regulation from both hosts, which enabled E. coli to function as whole-cell biosensor for CYA. To alleviate challenges of whole-cell biosensing, we adapted these promoter designs to function within a freeze-dried E. coli cell-free system to sense CYA. This portable, on-demand system robustly detects CYA within an hour from laboratory and real-world samples and works with both fluorescent and colorimetric reporters. This work elucidates general principles to facilitate the engineering of a wider array of LTTR-based environmental sensors.

synthetic biology↗

bioRxiv2018

Thermodynamic integration for dynamic causal models

In generative modeling of neuroimaging data, such as dynamic causal modeling (DCM), one typically considers several alternative models, either to determine the most plausible explanation for observed data (Bayesian model selection) or to account for model uncertainty (Bayesian model averaging). Both procedures rest on estimates of the model evidence, a principled trade-off between model accuracy and complexity. In DCM, the log evidence is usually approximated using variational Bayes (VB) under the Laplace approximation (VBL). Although this approach is highly efficient, it makes distributional assumptions and can be vulnerable to local extrema. An alternative to VBL is Markov Chain Monte Carlo (MCMC) sampling, which is asymptotically exact but orders of magnitude slower than VB. This has so far prevented its routine use for DCM.\n\nThis paper makes four contributions. First, we introduce a powerful MCMC scheme - thermodynamic integration (TI) - to neuroimaging and present a derivation that establishes a theoretical link to VB. Second, this derivation is based on a tutorial-like introduction to concepts of free energy in physics and statistics. Third, we present an implementation of TI for DCM that rests on population MCMC. Fourth, using simulations and empirical functional magnetic resonance imaging (fMRI) data, we compare log evidence estimates obtained by TI, VBL, and other MCMC-based estimators (prior arithmetic mean and posterior harmonic mean). We find that model comparison based on VBL gives reliable results in most cases, justifying its use in standard DCM for fMRI. Furthermore, we demonstrate that for complex and/or nonlinear models, TI may provide more robust estimates of the log evidence. Importantly, accurate estimates of the model evidence can be obtained with TI in acceptable computation time. This paves the way for using DCM in scenarios where the robustness of single-subject inference and model selection becomes paramount, such as differential diagnosis in clinical applications.

neuroscience↗