Biology subjects

Opuu, V.

Publications and source records attributed to Opuu, V..

9 recordsLinked to original sources

bioRxiv2026

DORA: a dose-response autoencoder for interpretable transcriptome-to-viability prediction

Predicting the effect of drugs on cell viability is a central challenge in drug discovery. Artificial intelligence holds the promise to considerably accelerate this process by leveraging rich cellular data such as transcriptomics. Current models focus on either transcriptomes or inhibitory concentrations, but they fall short in integrating these sources of information. Here, we propose DORA (Dose-Response Autoencoder), a deep learning model that predicts changes in transcriptomes and viability in a dose-dependent manner, knowing the unperturbed cell state. By enforcing a latent space consistent with cumulative dose effects, DORA matches other methods at predicting transcriptomes and substantially outperforms existing latent representations at viability prediction. The transcriptome-viability relationship provided by the model further allows the recovery of known biomarkers of cell viability while suggesting novel ones. Overall, DORA provides a unified framework delivering actionable biological insights for phenotypic drug screening and personalized medicine.

molecular biology↗

bioRxiv2025

Hybrid Generative Model: Bridging Machine Learning and Biophysics to Expand RNA Functional Diversity

Functional RNAs perform diverse catalytic roles, yet natural sequences represent only a narrow subset of what is possible. Rediscovering such activities requires exploring functional sequence diversity beyond natural RNAs. We introduce a hybrid generative model that combines a coevolutionary likelihood with an RNA secondary structure prior. This approach disentangles folding constraints from functional signals, enabling targeted diversification. On synthetic benchmarks, the model generates functional sequences beyond the training distribution. On large-scale ribozyme data, it improves the detection of active sequences and enhances sensitivity to local tertiary contacts. Finally, we introduce structural imprinting, a sampling strategy that uses alternative secondary structures to steer generation across under-sampled regions of sequence space. These results show that folding-informed generative modeling improves RNA design by supporting both extrapolation and control.

bioinformatics↗

bioRxiv2024

A Guide for Active Learning in Synergistic Drug Discovery

Synergistic drug combination screening is a promising strategy in drug discovery, but it involves navigating a costly and complex search space. While AI, particularly deep learning, has advanced synergy predictions, its effectiveness is limited by the low occurrence of synergistic drug pairs. Active learning, which integrates experimental testing into the learning process, has been proposed to address this challenge. In this work, we explore the key components of active learning to provide recommendations for its implementation. We find that molecular encoding has a limited impact on performance, while the cellular environment features significantly enhance predictions. Additionally, active learning can discover 60% of synergistic drug pairs with only exploring 10% of combinatorial space. The synergy yield ratio is observed to be even higher with smaller batch sizes, where dynamic tuning of the exploration-exploitation strategy can further enhance performance.

bioinformatics↗

bioRxiv2024

Expanding the space of self-reproducing RNAs using probabilistic generative models

Estimating the plausibility of RNA self-reproduction is central to origin-of-life scenarios but self-reproduction has been shown in only a handful of systems. Here, we populated a vast sequence space of ribozymes using statistical covariation models and secondary structure prediction. Experimentally assayed sequences were found active as far as 65 mutations from a reference natural sequence. The number of potentially generated sequences together with the experimental success rate indicate that at least [~]1039 such ribozymes may exist. Randomly sampled artificial ribozymes exhibited autocatalytic self-reproduction akin to the reference sequence. The combination of high-throughput screening and probabilistic modeling considerably improves our estimation of the number of self-reproducing systems, paving the way for a statistical approach to the origin of life.

biochemistry↗

bioRxiv2024

Hypothesis-driven interpretable neural network for interactions between genes

Mechanistic models of genetic interactions are rarely feasible due to a lack of information and computational challenges. Alternatively, machine learning (ML) approaches may predict gene interactions if provided with enough data but they lack interpretability. Here, we propose an ML approach for interpretable genotype-to-fitness mapping, the Direct-Latent Interpretable Model (D-LIM). The neural network is built on a strong hypothesis: mutations in different genes cause independent effects in phenotypes, which then interact via non-linear relationships to determine fitness. D-LIM predicts interpretable genotype-to-fitness maps with state-of-the-art accuracy for gene-to-gene and gene-to-environment perturbations in deep mutational scanning of a metabolic pathway, a protein-protein interaction system, and yeast mutants for environmental adaptation. The hypothesis-driven structure of D-LIM offers interpretable features reminiscent of mechanistic models: the inference of phenotypes, identification of trade-offs, and fitness extrapolation outside of the data domain.

bioinformatics↗

bioRxiv2022

Redesigning methionyl-tRNA synthetase for β-methionine activity with adaptive landscape flattening and experiments

Amino acids (AAs) with a noncanonical backbone would be a valuable tool for protein engineering, enabling new structural motifs and building blocks. To incorporate them into an expanded genetic code, the first, key step is to obtain an appropriate aminoacyl-tRNA synthetase (aaRS). Currently, directed evolution is not available to optimize such AAs, since an appropriate selective pressure is not available. Computational protein design (CPD) is an alternative. We used a new CPD method to redesign MetRS and increase its activity towards {beta}-Met, which has an extra backbone methylene. The new method considered a few active site positions for design and used a Monte Carlo exploration of the corresponding sequence space. During the exploration, a bias energy was adaptively learned, such that the free energy landscape of the apo enzyme was flattened. Enzyme variants could then be sampled, in the presence of the ligand and the bias energy, according to their {beta}-Met binding affinities. Eleven predicted variants were chosen for experimental testing; all exhibited detectable activity for {beta}-Met adenylation. Top predicted hits were characterized experimentally in detail. Dissociation constants, catalytic rates, and Michaelis constants for both -Met and {beta}-Met were measured. The best mutant retained a preference for -Met over {beta}-Met; however, the preference was reduced, compared to the wildtype, by a factor of 29. For this mutant, high resolution crystal structures were obtained in complex with both -Met and {beta}-Met, indicating that the predicted, active conformation of {beta}-Met in the active site was retained. Author summaryAmino acids (AAs) with a noncanonical backbone would be valuable for protein engineering, enabling new structural motifs. To incorporate them into an expanded genetic code, the key step is to obtain an appropriate aminoacyl-tRNA synthetase (aaRS). Currently, directed evolution is not available to optimize such AAs. Computational protein design is an alternative. We used a new method to redesign MetRS and increase its activity towards {beta}-Met, which has an extra backbone methylene. The method considered a few active site positions for design and used a Monte Carlo exploration of sequence space, during which a bias energy was adaptively learned, such that the free energy landscape of the apo enzyme was flattened. Enzyme variants could then be sampled, in the presence of the ligand and the bias energy, according to their {beta}-Met binding affinities. Eleven predicted variants were chosen for experimental testing; all exhibited detectable {beta}-Met adenylation activity. Top hits were characterized experimentally in detail. The best mutant had its preference for -Met over {beta}-Met reduced by a factor of 29. Crystal structures indicated that the predicted, active conformation of {beta}-Met in the active site was retained.

biochemistry↗

bioRxiv2020

Proteus software for physics-based protein design

We describe methods and software for physics-based protein design. The folded state energy combines molecular mechanics with Generalized Born solvent. Sequence and conformation space are sampled with Replica Exchange Monte Carlo, assuming one or a few fixed protein backbone structures and discrete side chain rotamers. Whole protein design and enzyme design are presented as illustrations. Full redesign of three PDZ domains was done using a simple, empirical, unfolded state model. Designed sequences were very similar to natural ones. Enzyme redesign exploited a powerful, adaptive, importance sampling approach that allows the design to directly target substrate binding, reaction rate, catalytic efficiency, or the specificity of these properties. Redesign of tyrosyl-tRNA synthetase stereospecificity is reported as an example.Competing Interest StatementThe authors have declared no competing interest.View Full Text

biophysics↗

bioRxiv2019

A physics-based energy function allows the computational redesign of a PDZ domain

A powerful approach to understand protein structure and evolution is to perform computer simulations that mimic aspects of evolution. In particular, structure-based computational protein design (CPD) can address the inverse folding problem, exploring a large space of amino acid sequences and selecting ones predicted to adopt a given fold. Previously, CPD has been used to entirely redesign several proteins: all or most of the protein sequence was allowed to mutate freely; among sampled sequences, those with low computed folding energy were selected, and a few percent of them did indeed adopt the correct fold. Those studies used an energy function that was partly or largely knowledge-based, with several empirical terms. Here, we show that a PDZ domain can be entirely redesigned using a \"physics-based\" energy function that combines standard molecular mechanics and a recent, continuum electrostatic solvent model. Many thousands of sequences were generated by Monte Carlo simulation. Among the lowest-energy sequences, three were chosen for experimental testing. All three could be overexpressed and had native-like circular dichroism and 1D-NMR spectra. Two exhibited an increase in their thermal denaturation curves when a peptide ligand was present, indicating binding and suggesting correctly folded proteins. Evidently, the physical principles that govern molecular mechanics and continuum electrostatics are sufficient to perform whole-protein redesign. This is encouraging, since these methods provide physical insights, can be systematically improved, and are transferable to other biopolymers and ligands of medical or technological interest.\n\n\n\nO_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=73 SRC=\"FIGDIR/small/790667v3_ufig1.gif\" ALT=\"Figure 1\">\nView larger version (26K):\norg.highwire.dtl.DTLVardef@145ea28org.highwire.dtl.DTLVardef@10ffa2forg.highwire.dtl.DTLVardef@1f2460eorg.highwire.dtl.DTLVardef@d5691c_HPS_FORMAT_FIGEXP M_FIG Table of Contents Graphic\n\nC_FIG

biophysics↗

bioRxiv2019

Adaptive landscape flattening allows the design of both enzyme:substrate binding and catalytic power

Designed enzymes are of fundamental and technological interest. Experimental directed evolution still has significant limitations, and computational approaches are complementary. A designed enzyme should satisfy multiple criteria: stability, substrate binding, transition state binding. Such multi-objective design is computationally challenging. Two recent studies used adaptive importance sampling Monte Carlo to redesign proteins for ligand binding. By first flattening the energy landscape of the apo protein, they obtained positive design for the bound state and negative design for the unbound. We extend the method to the design of an enzyme for specific transition state binding, i.e., for catalytic power. We consider methionyl-tRNA synthetase (MetRS), which attaches methionine (Met) to its cognate tRNA, establishing codon identity. MetRS and other synthetases have been extensively redesigned by experimental directed evolution to accept noncanonical amino acids as substrates, leading to genetic code expansion. We redesigned MetRS computationally to bind several ligands: the Met analog azidonorleucine, methionyl-adenylate (MetAMP), and the activated ligands that form the transition state for MetAMP production. Enzyme mutants known to have azidonorleucine activity were recovered, and mutants predicted to bind MetAMP were characterized experimentally and found to be active. Mutants predicted to have low activation free energies for MetAMP production were found to be active and the predicted reaction rates agreed well with the experimental values. We expect the present method will become the paradigm for computational enzyme design.

biochemistry↗