Biology subjects

Monasson, R.

Publications and source records attributed to Monasson, R..

35 records · Page 2Linked to original sources

bioRxiv2023

Computational design of novel Cas9 PAM-interacting domains using evolution-based modelling and structural quality assessment

We present here an approach to protein design that enables to leverage (i) scarce functional information such as experimental data (ii) evolutionary information learned from a natural sequence variants and (iii) physics-grounded modeling. Using a Restricted Boltzmann Machine (RBM), we learned a sequence model of a protein family. We use semi-supervision to leverage available functional information during the RBM training. We then propose a strategy to explore the protein representation space that can be informed by external models such as an empirical force field method (FoldX). This method was applied to a domain of the Cas9 protein responsible for recognition of a short DNA motif. We experimentally assessed the functionality of 71 variants that were generated to explore a range of RBM and FoldX energies. We show how a combination of functional, structural and evolutionary information can identify functional variants with high accuracy. Sequences with as many as 50 differences (20% of the protein domain) to the wild-type retained functionality. Overall, 21/71 sequences designed with our method were functional. Interestingly, 6/71 sequences showed an improved activity in comparison with the original wild-type protein sequence. These results demonstrate the interest in further exploring the synergies between machine-learning of protein sequence representations and physics grounded modeling strategies informed by structural information.

bioinformatics↗

bioRxiv2022

Infer global, predict local: quantity-quality trade-off in protein fitness predictions from sequence data

Predicting the effects of mutations on protein function is an important issue in evolutionary biology and biomedical applications. Computational approaches, ranging from graphical models to deep-learning architectures, can capture the statistical properties of sequence data and predict the outcome of high-throughput mutagenesis experiments probing the fitness landscape around some wild-type protein. However, how the complexity of the models and the characteristics of the data combine to determine the predictive performance remains unclear. Here, based on a theoretical analysis of the prediction error, we propose descriptors of the sequence data, characterizing their quantity and quality relative to the model. Our theoretical framework identifies a trade-off between these two quantities, and determines the optimal subset of data for the prediction task, showing that simple models can outperform complex ones when inferred from adequately-selected sequences. We also show how repeated subsampling of the sequence data allows for assessing how much epistasis in the fitness landscape is not captured by the computational model. Our approach is illustrated on several protein families, as well as on in silico solvable protein models. Significance StatementIs more data always better? Or should one prefer fewer data, but of higher quality? Here, we investigate this question in the context of the prediction of fitness effects resulting from mutations to a wild-type protein. We show, based on theory and data analysis, that simple models trained on a small subset of carefully chosen sequence data can perform better than complex ones trained on all available data. Furthermore, we explain how comparing the simple local models obtained with different subsets of training data reveals how much of the epistatic interactions shaping the fitness landscape are left unmodeled.

biophysics↗

bioRxiv2022

Learning the differences: a transfer-learning approach to predict antigen immunogenicity and T-cell receptor specificity

Antigen immunogenicity and the specificity of binding of T-cell receptors to antigens are key properties underlying effective immune responses. Here we propose diffRBM, an approach based on transfer learning and Restricted Boltzmann Machines, to build sequence-based predictive models of these properties. DiffRBM is designed to learn the distinctive patterns in amino acid composition that, one the one hand, underlie the antigens probability of triggering a response, and on the other hand the T-cell receptors ability to bind to a given antigen. We show that the patterns learnt by diffRBM allow us to predict putative contact sites of the antigen-receptor complex. We also discriminate immunogenic and non-immunogenic antigens, antigen-specific and generic receptors, reaching performances that compare favorably to existing sequence-based predictors of antigen immunogenicity and T-cell receptor specificity. More broadly, diffRBM provides a general framework to detect, interpret and leverage selected features in biological data.

bioinformatics↗

bioRxiv2022

Generative and interpretable machine learning for aptamer design and analysis of in vitro sequence selection

Selection protocols such as SELEX, where molecules are selected over multiple rounds for their ability to bind to a target molecule of interest, are popular methods for obtaining binders for diagnostic and therapeutic purposes. With the increasing amount of such high-throughput experimental data available, machine learning techniques have become increasingly popular for molecular datasets analysis. Here, we show that Restricted Boltzmann Machines (RBMs), a two-layer neural network architecture, can successfully be trained on sequence ensembles from SELEX experiments for thrombin aptamers, and used to estimate the fitness of the sequences obtained through the experimental protocol. As a direct consequence, we show that trained RBMs can be exploited to classify as well as generate novel molecules. To confirm our findings, we experimentally verify the generated sequences from RBM.

biophysics↗

bioRxiv2022

Emergence of time persistence in an interpretable data-driven neural network model

Establishing accurate as well as interpretable models of network activity is an open challenge in systems neuroscience. Here we infer an energy-based model of the ARTR, a circuit that controls zebrafish swimming statistics, using functional recordings of the spontaneous activity of hundreds of neurons. Although our model is trained to reproduce the low-order statistics of the network activity at short time-scales, its simulated dynamics quantitatively captures the slowly alternating activity of the ARTR. It further reproduces the modulation of this persistent dynamics by the water temperature and visual stimulation. Mathematical analysis of the model unveils a low-dimensional landscape-based representation of the ARTR activity, where the slow network dynamics reflects Arrhenius-like barriers crossings between metastable states. Our work thus shows how data-driven models built from large neural populations recordings can be reduced to low-dimensional functional models in order to reveal the fundamental mechanisms controlling the collective neuronal dynamics.

neuroscience↗

bioRxiv2021

Origins and breadth of pairwise epistasis in an α-helix of β-lactamase TEM-1

Epistasis affects genome evolution together with our ability to predict individual mutation effects. The mechanistic basis of epistasis remains, however, largely unknown. To quantify and better understand interactions between fitness-affecting mutations, we focus on a 11 amino-acid -helix of the protein {beta}-lactamase TEM-1, and build a comprehensive library of more than 15,000 double mutants. Analysis of the growth rates of these mutants shows pervasive epistasis, which can be largely explained by a non-linear two-state model, where inactivating, destabilizing, neutral, or stabilizing mutations additively contribute to the phenotype. Hence, most epistatic interactions can be predicted by a non-linear model informed by single-point mutational measurements only. Deviations from the two-state model are consistently found for few pairs of residues, in particular when they are in contact. This result, as well as single-point mutation parameters, can be quantitatively found back through direct-coupling-analysis-based statistical models inferred from homologous sequence data. Our results thus shed light on the existence and the origins of the multiple determinants of the epistatic landscape, even at the level of small structural components of a protein, and suggest that the corresponding constraints shape the entire {beta}-lactamase family.

evolutionary biology↗

bioRxiv2021

Repeats Mimic Immunostimulatory Viral Features Across a Vast Evolutionary Landscape

An emerging hallmark across human diseases - such as cancer, autoimmune and neurodegenerative disorders - is the aberrant transcription of typically silenced repetitive elements. Once active, a subset of repeats may be capable of "viral mimicry": the display of pathogen-associated molecular patterns (PAMPs) that can, in principle, bind pattern recognition receptors (PRRs) of the innate immune system and trigger inflammation. Yet how to quantify the landscape of viral mimicry and how it is shaped by natural selection remains a critical gap in our understanding of both genome evolution and the immunological basis of disease. We propose a theoretical framework to quantify selective forces on virus-like features as the entropic cost a sequence pays to hold a non-self PAMP and show our approach can predict classes of viral-mimicry within the human genome and across eukaryotes. We quantify the breadth and conservation of viral mimicry across multiple species for the first time and integrate selective forces into predictive evolutionary models. We show HSATII and intact LINE-1 (L1) are under selection to maintain CpG motifs, and specific Alu families likewise maintain the proximal presence of inverted copies to form double-stranded RNA (dsRNA). We validate our approach by predicting high CpG L1 ligands of L1 proteins and the innate receptor ZCCHC3, and dsRNA present both intracellularly and as MDA5 ligands. We conclude viral mimicry is a general evolutionary mechanism whereby genomes co-opt pathogen-associated features generated by prone repetitive sequences, likely offering an advantage as a quality control system against transcriptional dysregulation.

systems biology↗

bioRxiv2021

A synaptic novelty signal in the dentate gyrus supports switching hippocampal attractor networks from generalization to discrimination

Episodic memory formation and recall are complementary processes that put conflicting requirements on neuronal computations in the hippocampus. How this challenge is resolved in hippocampal circuits is unclear. To address this question, we obtained in vivo whole-cell patch-clamp recordings from dentate gyrus granule cells in head-fixed mice trained to explore and distinguish between familiar and novel virtual environments. We find that granule cells consistently show a small transient depolarization of their membrane potential upon transition to a novel environment. This synaptic novelty signal is sensitive to local application of atropine, indicating that it depends on metabotropic acetylcholine receptors. A computational model suggests that the observed transient synaptic response to novel environments may lead to a bias in the granule cell population activity, which can in turn drive the downstream attractor networks to a new state, thereby favoring the switch from generalization to discrimination when faced with novelty. Such a novelty-driven cholinergic switch may enable flexible encoding of new memories while preserving stable retrieval of familiar ones.

neuroscience↗

bioRxiv2021

Improving sequence-based modeling of protein families using secondary structure quality assessment

MotivationModeling of protein family sequence distribution from homologous sequence data recently received considerable attention, in particular for structure and function predictions, as well as for protein design. In particular, Direct Coupling Analysis, a method to infer effective pairwise interactions between residues, was shown to capture important structural constraints and to successfully generate functional protein sequences. Building on this and other graphical models, we introduce a new framework to assess the quality of the secondary structures of the generated sequences with respect to reference structures for the family. ResultsWe introduce two scoring functions characterizing the likeliness of the secondary structure of a protein sequence to match a reference structure, called Dot Product and Pattern Matching. We test these scores on published experimental protein mutagenesis and design dataset, and show improvement in the detection of non-functional sequences. We also show that use of these scores help rejecting non-functional sequences generated by graphical models (Restricted Boltzmann Machines) learned from homologous sequence alignments. AvailabilitySupplementary Materials, Data and Code available at https://github.com/CyrilMa/ssqa. Contactcyril.malbranke@ens.fr O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=141 SRC="FIGDIR/small/428964v1_ufig1.gif" ALT="Figure 1"> View larger version (38K): org.highwire.dtl.DTLVardef@38d327org.highwire.dtl.DTLVardef@aeba12org.highwire.dtl.DTLVardef@13a09a2org.highwire.dtl.DTLVardef@14e85fa_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2020

Probing T-cell response by sequence-based probabilistic modeling

With the increasing ability to use high-throughput next-generation sequencing to quantify the diversity of the human T cell receptor (TCR) repertoire, the ability to use TCR sequences to infer antigen-specificity could greatly aid potential diagnostics and therapeutics. Here, we use a machine-learning approach known as Restricted Boltzmann Machine to develop a sequence-based inference approach to identify antigen-specific TCRs. Our approach combines probabilistic models of TCR sequences with clone abundance information to extract TCR sequence motifs central to an antigen-specific response. We use this model to identify patient personalized TCR motifs that respond to individual tumor and infectious disease antigens, and to accurately discriminate specific from non-specific responses. Furthermore, the hidden structure of the model results in an interpretable representation space where TCRs responding to the same antigen cluster, correctly discriminating the response of TCR to different viral epitopes. The model can be used to identify condition specific responding TCRs. We focus on the examples of TCRs reactive to candidate neoantigens and selected epitopes in experiments of stimulated TCR clone expansion. Author SummaryLarge repertoires of immune cells, such as T cells, are increasingly made available by high-throughput sequencing. Exploiting such datasets to infer how T-cell respond to antigens could help design vaccines and adoptive T-cell therapies. We here propose an approach based on probabilistic machine learning to identify and characterize responding T cells. After learning, this approach is able to distinguish clones that specifically respond to different antigen stimulations. The model parameters and the low-dimensional representations of the T-cell sequences identify sequence motifs underlying T-cell recognition at the molecular level. The approach is illustrated on repertoire data describing in vitro stimulation of T cells by cancer-related neoantigens, as well as on data for common infectious diseases.

bioinformatics↗

bioRxiv2020

Quantitative modeling of the effect of antigen dosage on B-cell affinity distributions in maturating germinal centers

Affinity maturation is a complex dynamical process allowing the immune system to generate antibodies capable of recognizing antigens. We introduce a model for the evolution of the distribution of affinities across the antibody population in germinal centers. The model is amenable to detailed mathematical analysis, and gives insight on the mechanisms through which antigen availability controls the rate of maturation and the expansion of the antibody population. It is also capable, upon maximum-likelihood inference of the parameters, to reproduce accurately the distributions of affinities of IgG-secreting cells we measure in mice immunized against Tetanus Toxoid under largely varying conditions (antigen dosage, delay between injections). Both model and experiments show that the average population affinity depends non-monotonically on the antigen dosage. We show that combining quantitative modelling and statistical inference is a concrete way to investigate biological processes underlying affinity maturation (such as selection permissiveness), hardly accessible through measurements.

immunology↗

bioRxiv2020

The heterogeneous landscape and early evolution of pathogen-associated CpG and UpA dinucleotides in SARS CoV-2

COVID-19 can lead to acute respiratory syndrome, which can be due to dysregulated immune signaling. We analyze the distribution of CpG dinucleotides, a pathogen-associated molecular pattern, in the SARS-CoV-2 genome. We find that the CpG content, which we characterize by a force parameter that accounts for statistical constraints acting on the genome at the nucleotidic and amino-acid levels, is, on average, low compared to other pathogenic betacoronaviruses. However, the CpG force widely fluctuates along the genome, with a particularly low value, comparable to the circulating seasonal HKU1, in the spike coding region and a greater value, comparable to SARS and MERS, in the highly expressed nucleocapside coding region (N ORF), whose transcripts are relatively abundant in the cytoplasm of infected cells and present in the 3UTRs of all subgenomic RNA. This dual nature of CpG content could confer to SARS-CoV-2 the ability to avoid triggering pattern recognition receptors upon entry, while eliciting a stronger response during replication. We then investigate the evolution of synonymous mutations since the outbreak of the COVID-19 pandemic, finding a signature of CpG loss in regions with a greater CpG force. Sequence motifs preceding the CpG-loss-associated loci in the N ORF match recently identified binding patterns of the Zinc finger Anti-viral Protein. Using a model of the viral gene evolution under human host pressure, we find that synonymous mutations seem driven in the SARS-CoV-2 genome, and particularly in the N ORF, by the viral codon bias, the transition-transversion bias and the pressure to lower CpG content.

genomics↗

bioRxiv2020

Flexible machine learning prediction of antigen presentation for rare and common HLA-I alleles

The recent increase of immunopeptidomic data, obtained by mass spectrometry or binding assays, opens unprecedented possibilities for investigating endogenous antigen presentation by the highly polymorphic human leukocyte antigen class I (HLA-I) protein. We introduce a flexible and easily interpretable peptide presentation prediction method, RBM-MHC. We validate its performance as a predictor of cancer neoantigens and viral epitopes and we use it to reconstruct peptide motifs presented on specific HLA-I molecules. By benchmarking RBM-MHC performance on a wide range of HLA-I alleles, we show its importance to improve prediction accuracy for rarer alleles.

bioinformatics↗

bioRxiv2020

Evolution-based design of chorismate mutase enzymes

The rational design of enzymes is an important goal for both fundamental and practical reasons. Here, we describe a design process in which we learn the constraints for specifying proteins purely from evolutionary sequence data, build libraries of synthetic genes, and test them for activity in vivo using a quantitative complementation assay. For chorismate mutase, a key enzyme in the biosynthesis of aromatic amino acids, we demonstrate the design of natural-like catalytic function with substantial sequence diversity. Further optimization focuses the generative model towards function in a specific genomic context. The data show that sequence-based statistical models suffice to specify proteins and provide access to an enormous space of synthetic functional sequences. This result provides a foundation for a general process for evolution-based design of artificial proteins. One-sentence summaryAn evolution-based, data-driven engineering process can build synthetic functional enzymes.

biochemistry↗

bioRxiv2019

Parameters and determinants of responses to selection in antibody libraries

Antibody repertoires contain binders to nearly any target antigen. The sequences of these antibodies differ mostly at few sites located on the surface of a scaffold that itself consists of much less varied amino acids. What is the impact of this scaffold on the response to selection of a repertoire? To gauge this impact, we carried out quantitative phage display experiments with three antibody libraries based on distinct scaffolds harboring the same diversity at randomized sites, which we selected for binding to four arbitrary targets. We first show that the response to selection of an antibody library is captured by a simple and measurable parameter with direct physical and information-theoretic interpretations. Second, we identify a major determinant of this parameter which is encoded in the scaffold, its degree of evolutionary maturation. Antibodies undergo an accelerated evolutionary process, called affinity maturation, to improve their affinity to a given target antigen as part of the adaptive immune response. We find that libraries of antibodies built around such maturated scaffolds have a lower response to selection to other arbitrary targets than libraries built around naive scaffolds of germline origin. Our results are a first step towards quantifying and controlling the evolutionary potential of biomolecules.

biophysics↗

bioRxiv2019

Can grid cell ensembles represent multiple spaces?

The way grid cells represent space in the rodent brain has been a striking discovery, with theoret-ical implications still unclear. Differently from hippocampal place cells, which are known to encode multiple, environment-dependent spatial maps, grid cells have been widely believed to encode space through a single low dimensional manifold, in which coactivity relations between different neurons are preserved when the environment is changed. Does it have to be so? Here, we compute - using two alternative mathematical models - the storage capacity of a population of grid-like units, em-bedded in a continuous attractor neural network, for multiple spatial maps. We show that distinct representations of multiple environments can coexist, as existing models for grid cells have the po-tential to express several sets of hexagonal grid patterns, challenging the view of a universal grid map. This suggests that a population of grid cells can encode multiple non-congruent metric rela-tionships, a feature that could in principle allow a grid-like code to represent environments with a variety of different geometries and possibly conceptual and cognitive spaces, which may be expected to entail such context-dependent metric relationships.

neuroscience↗

bioRxiv2018

Environmental adaptation of olfactory receptor distributions

Olfactory receptor usage is highly non-uniform, with some receptor types being orders of magnitude more abundant than others. We propose an explanation for this striking fact: the receptor distribution is tuned to maximally represent information about the olfactory environment in a little-studied regime of efficient coding that is sensitive to the global context of correlated sensor responses. This model predicts further that in mammals, where olfactory sensory neurons are replaced regularly, receptor abundances should continuously adapt to odor statistics. Indeed, experiments have found such changes in response to olfactory experience. These changes are mysteriously context-dependent, with increased exposure to odorants leading variously to increased, decreased or unchanged abundances of activated receptors. We show that efficient coding theory predicts precisely such a context-dependence when sensors are correlated. Finally, we demonstrate simple dynamical rules for neural birth and death processes that might provide a mechanism for receptor abundances to adapt optimally.

neuroscience↗