Biology subjects

Petti, S.

Publications and source records attributed to Petti, S..

10 recordsLinked to original sources

bioRxiv2026

Inferring modularity in genotype-phenotype maps using patterns of epistasis and pleiotropy

Methods for inferring the relationship between genotype and phenotype often search for low-dimensional structure, motivated by the hypothesis that genetic loci can be organized into a small number of groups representing biological processes. These methods typically group genetic loci based on which other loci they interact with (epistasis) or based on how their effects covary across environmental conditions (patterns of pleiotropy). Here, we present two complementary methods that combine epistasis and pleiotropy to infer modular structure. The first, Sparse, Low-dimensional, Epistatic Decomposition (SLED), parametrizes the genotype-phenotype map in terms of a small number of modules: loci contribute sparse additive and epistatic effects to modules, and modules contribute sparse additive and epistatic effects to multiple phenotypes. The second method, Graph Tensor Decomposition (GTD), represents the epistatic interactions in each environment as a combination of shared basis graphs, each capturing a pattern of epistasis that recurs across environments. Since GTD does not use additive effects, learned graphs are not restricted to include only loci whose additive effects covary across environments. We validate SLED and GTD on synthetic data, then apply both methods to an experimental dataset consisting of the growth rates of yeast segregants from an F1 cross, measured across 101 environments with a hub-and-spoke structure. We evaluate the biological structure of the inferred modules and show that both models generalize well to held-out environments. Finally, we compare the modules inferred by SLED, GTD, and a previously established method that infers sparse modular structure from a locus-by-environment matrix of additive effects (SSD) and find that all three methods yield highly similar modular architectures. This suggests that the genetic architecture underlying this dataset is consistent across additive effects (SSD), epistatic effects (GTD), and both jointly (SLED). Together, our methods provide a framework for combining two sources of information, epistasis and pleiotropy, to infer interpretable low-dimensional structure in genotype-phenotype maps, expanding on previous work that has traditionally treated each independently.

evolutionary biology↗

bioRxiv2026

Joint QTL mapping reveals patterns of pleiotropy across structured environments in budding yeast

Quantitative trait locus (QTL) mapping studies in model organisms have long been used to infer the structure of genotype-phenotype maps. However, prior work has largely focused on analyzing the genetic architecture of complex traits individually. Here, we sought to instead infer genotype-phenotype maps for a large set of related phenotypes simultaneously. To do so, we used bulk assays to measure the competitive fitness of each of 100,000 individuals from a large budding yeast QTL mapping panel across a set of 101 phenotypes related by a ``hub-and-perturbation'' pattern, and then used a forward search approach to jointly infer the genetic basis of all phenotypes together. Our results reveal complex patterns of pleiotropy in the genetic architecture of these traits. We find that these patterns of pleiotropy reflect the structure of the environmental relationships as well as functional information about individual variants and the interactions between them. This suggests that high-throughput QTL mapping across structured environments can help uncover biologically meaningful genetic interactions, along with their variation across related environmental contexts.

evolutionary biology↗

bioRxiv2025

Learning sequence-function relationships with scalable, interpretable Gaussian processes

AbstractUnderstanding the relationship between biological sequences, such as DNA, RNA or protein sequences, and their resulting phenotypes is one of the central goals of genetics. This task is complicated by epistasis, i.e., the context dependence of mutational effects. Advances in high-throughput phenotyping now make it possible to study these relationships at unprecedented scale, generating large datasets that measure phenotypes for tens or hundreds of thousands of sequences. However, standard regression models for analyzing such datasets often make unrealistic assumptions about the generalizability of mutational effects and epistatic coefficients across genetic backgrounds. Deep neural networks offer greater flexibility but suffer from limited interpretability and lack uncertainty quantification. Here, we introduce a family of interpretable Gaussian process models for sequence-function relationships that capture epistasis through flexible prior distributions that generalize classical theoretical models from the fitness landscape literature. In particular, these priors are parameterized by interpretable site-, allele-, and mutation-specific factors controlling the degree to which specific mutations decrease the predictability of the effects of other mutations. Using GPU acceleration to scale to large protein, RNA, and genome-wide SNP datasets, our models consistently deliver superior predictive performance while yielding interpretable parameters that both recover known features and uncover novel epistatic interactions. Overall, our methods provide new insights into the structure of the genotype-phenotype map and offer scalable, interpretable approaches for exploring complex genetic interactions across diverse biological systems.

systems biology↗

bioRxiv2025

SoftAlign: End-to-end protein structures alignment

With the recent breakthrough of highly accurate structure prediction methods, there has been a rapid growth of available protein structures. Efficient methods are needed to infer structural similarity within these datasets. We present an end-to-end alignment method, called SoftAlign, which takes as input the 3D coordinates of a protein pair and outputs a structural alignment. In addition to the traditional Smith-Waterman alignment method, we introduce a modified softmax alignment that shows very promising results for structure similarity detection. We demonstrate that the SoftAlign model is able to recapitulate TM-align alignments while running faster, and it is more accurate than Foldseek on alignment and classification tasks. Although SoftAlign is not the fastest method available, it is highly precise and can be used effectively with other prefilters. In addition to developing an end-to-end structural aligner, our main contribution is the introduction and analysis of a pseudo-alignment method based on softmax, which can be used with other architectures, even those not based on structural information. The code for SoftAlign is available at https://github.com/jtrinquier/SoftAlign.

bioinformatics↗

bioRxiv2025

On learning functions over biological sequence space: relating Gaussian process priors, regularization, and gauge fixing

Mappings from biological sequences (DNA, RNA, protein) to quantitative measures of sequence functionality play an important role in contemporary biology. We are interested in the related tasks of (i) inferring predictive sequence-to-function maps and (ii) decomposing sequence-function maps to elucidate the contributions of individual subsequences. Because each sequence-function map can be written as a weighted sum over subsequences in multiple ways, meaningfully interpreting these weights requires "gauge-fixing," i.e., defining a unique representation for each map. Recent work has established that most existing gauge-fixed representations arise as the unique solutions to L2-regularized regression in an overparameterized "weight space" where the choice of regularizer defines the gauge. Here, we establish the relationship between regularized regression in overparameterized weight space and Gaussian process approaches that operate in "function space," i.e. the space of all real-valued functions on a finite set of sequences. We disentangle how weight space regularizers both impose an implicit prior on the learned function and restrict the optimal weights to a particular gauge. We show how to construct regularizers that correspond to arbitrary explicit Gaussian process priors combined with a wide variety of gauges and characterize the implicit function space priors associated with the most common weight space regularizers. Finally, we derive the posterior distribution of a broad class of sequence-to-function statistics, including gauge-fixed weights and multiple systems for expressing higher-order epistatic coefficients. We show that such distributions can be efficiently computed for product-kernel priors using a kernel trick.

bioinformatics↗

bioRxiv2025

An interpretable alphabet for local protein structure search based on amino acid neighborhoods

MotivationRecent advancements in protein structure prediction methods have vastly increased the size of databases of protein structures, necessitating fast methods for protein structure comparison. Search methods that find structurally similar proteins can be applied to find remote homologs, study the functional relationships among proteins, and aid in protein engineering tasks. The structure comparison method Foldseek represents each protein structure as a sequence of "3Di" characters and uses highly optimized sequence comparison software to search with this alphabet. An alternate alphabet encoding richer features has the potential to improve search accuracy while leaving the underlying search algorithm unchanged. ResultsWe design a "3Dn" structural alphabet that encodes the local neighborhoods around each amino acid in an interpretable way. In a search benchmark task, a combination of our alphabet and Foldseeks 3Di alphabet, outperforms each alphabet individually and ranks best among local search methods that do not require amino acid identity information. We provide software tools that enable the exploration of novel alphabets and combinations of alphabets for protein structure search. Availability and implementationThe code is freely available at https://github.com/spetti/structure_comparison.

bioinformatics↗

bioRxiv2025

Inferring genotype-phenotype maps using attention models

Predicting phenotype from genotype is a central challenge in genetics. Traditional approaches in quantitative genetics typically analyze this problem using methods based on linear regression. These methods generally assume that the genetic architecture of complex traits can be parameterized in terms of an additive model, where the effects of loci are independent, plus (in some cases) pair-wise epistatic interactions between loci. However, these models struggle to analyze more complex patterns of epistasis or subtle gene-environment interactions. Recent advances in machine learning, particularly attention-based models, offer a promising alternative. Initially developed for natural language processing, attention-based models excel at capturing context-dependent interactions and have shown exceptional performance in predicting protein structure and function. Here, we apply attention-based models to quantitative genetics. We analyze the performance of this attention-based approach in predicting phenotype from genotype using simulated data across a range of models with increasing epistatic complexity, and using experimental data from a recent quantitative trait locus mapping study in budding yeast. We find that our model demonstrates superior out-of-sample predictions in epistatic regimes compared to standard methods. We also explore a more general multi-environment attention-based model to jointly analyze genotype-phenotype maps across multiple environments and show that such architectures can be used for "transfer learning" - predicting phenotypes in novel environments with limited training data.

genetics↗

bioRxiv2022

Inferring sparse structure in genotype-phenotype maps

Correlation among multiple phenotypes across related individuals may reflect some pattern of shared genetic architecture: individual genetic loci affect multiple phenotypes (an effect known as pleiotropy), creating observable relationships between phenotypes. A natural hypothesis is that pleiotropic effects reflect a relatively small set of common "core" cellular processes: each genetic locus affects one or a few core processes, and these core processes in turn determine the observed phenotypes. Here, we propose a method to infer such structure in genotype-phenotype data. Our approach, sparse structure discovery (SSD) is based on a penalized matrix decomposition designed to identify latent structure that is low-dimensional (many fewer core processes than phenotypes and genetic loci), locus-sparse (each locus affects few core processes) and/or phenotype-sparse (each phenotype is influenced by few core processes). Our use of sparsity as a guide in the matrix decomposition is motivated by the results of a novel empirical test indicating evidence of sparse structure in several recent genotype-phenotype data sets. First, we use synthetic data to show that our SSD approach can accurately recover core processes if each genetic locus affects few core processes or if each phenotype is affected by few core processes. Next, we apply the method to three datasets spanning adaptive mutations in yeast, genotoxin robustness assay in human cell lines, and genetic loci identified from a yeast cross, and evaluate the biological plausibility of the core process identified. More generally, we propose sparsity as a guiding prior for resolving latent structure in empirical genotype-phenotype maps.

evolutionary biology↗

bioRxiv2021

End-to-end learning of multiple sequence alignments with differentiable Smith-Waterman

Multiple Sequence Alignments (MSAs) of homologous sequences contain information on structural and functional constraints and their evolutionary histories. Despite their importance for many downstream tasks, such as structure prediction, MSA generation is often treated as a separate pre-processing step, without any guidance from the application it will be used for. Here, we implement a smooth and differentiable version of the Smith-Waterman pairwise alignment algorithm that enables jointly learning an MSA and a downstream machine learning system in an end-to-end fashion. To demonstrate its utility, we introduce SMURF (Smooth Markov Unaligned Random Field), a new method that jointly learns an alignment and the parameters of a Markov Random Field for unsupervised contact prediction. We find that SMURF learns MSAs that mildly improve contact prediction on a diverse set of protein and RNA families. As a proof of concept, we demonstrate that by connecting our differentiable alignment module to AlphaFold and maximizing predicted confidence, we can learn MSAs that improve structure predictions over the initial MSAs. Interestingly, the alignments that improve AlphaFold predictions are self-inconsistent and can be viewed as adversarial. This work highlights the potential of differentiable dynamic programming to improve neural network pipelines that rely on an alignment and the potential dangers of relying on black-box methods for optimizing predictions of protein sequences.

bioinformatics↗

bioRxiv2021

Constructing benchmark test sets for biological sequence analysis using independent set algorithms

Statistical inference and machine learning methods are benchmarked on test data independent of the data used to train the method. Biological sequence families are highly non-independent because they are related by evolution, so the strategy for splitting data into separate training and test sets is a nontrivial choice in benchmarking sequence analysis methods. A random split is insufficient because it will yield test sequences that are closely related or even identical to training sequences. Adapting ideas from independent set graph algorithms, we describe two new methods for splitting sequence data into dissimilar training and test sets. These algorithms input a sequence family and produce a split in which each test sequence is less than p% identical to any individual training sequence. These algorithms successfully split more families than a previous approach, enabling construction of more diverse benchmark datasets.

bioinformatics↗