Biology subjects

Ivankov, D. N.

Publications and source records attributed to Ivankov, D. N..

16 recordsLinked to original sources

bioRxiv2026

Blockiness as Determinant of Liquid Phase Formation by Polypeptide-Like Polymers in Stickers-Spacers Paradigm

Liquid-liquid phase separation (LLPS) driven by intrinsically disordered protein regions underlies the formation of membraneless organelles. The role of these structures in cell biology is an active area of research, yet their physical understanding remains less developed than that of protein assemblies with well-defined, rigid tertiary structures. The amino acid composition of disordered regions, particularly the abundance of charged residues, is known to modulate phase behavior. These factors govern accessible conformations, the assembly of compact sticker clusters, and the porosity of the resulting polymer networks. In contrast, the spatial distribution of side-chain volumes remains a comparatively understudied sequence determinant. Here, we employ a model that captures the patterning of side-chain volumes and interaction strengths without specifying chemical identity to show how sequence blockiness modulates the phase-nucleation propensity, connectivity, and compactness of polypeptide-like polymer assemblies. Using a library of polymers with identical composition but permuted residue arrangements, we show that clustering of bulky residues generates cooperative sticker domains and identify the threshold blockiness required to stabilize polymer networks. Coarse-grained simulations and mean-field theory reveal that this effect arises from an interplay among steric repulsion, solvent entropy, and higher-valency interactions, whereby blocky patterns promote the formation of multi-node percolated networks. Segregation of stickers and spacers is therefore one of the sequence parameters governing LLPS which may be useful for interpreting the evolution of disordered protein sequences, high-throughput proteome screening, and engineering of stimuli-responsive biomaterials. These rules may be exploited to predetermine condensate stability, viscoelasticity, and molecular exchange rates without altering chemical functionality.

biophysics↗

bioRxiv2026

Hairpin Architecture and Extended Sequence Context Shape APOBEC3 Mutagenesis in Monkeypox Virus at Different Genomic Scales

The 2022 global mpox outbreak was characterized by an unusual accumulation of more than fifty APOBEC-signature substitutions in MPXV genomes. A previous study suggested that these substitutions preferentially occur within and near hairpins or cruciforms - stem-loop structures formed by inverted repeats. Here, we show that this conclusion resulted from an inappropriate statistical analysis and, using two substantially larger datasets, demonstrate that APOBEC-signature mutations are specifically enriched at TC motifs located at the 3' ends of hairpin loops, but not in other regions of hairpins/cruciforms or in their vicinity. We further identified a preference for stem-loop structures with longer stems and shorter loops, consistent with patterns previously reported for the APOBEC mutagenesis in cancer genomes. Our analysis also revealed a positive association between APOBEC-signature mutation propensity and hairpin stability in single-stranded DNA. Nevertheless, stem-loop structures with long stems and short loops are relatively rare in the MPXV genome, limiting their contribution to the genome-wide distribution of APOBEC-signature mutations to a predominantly local effect. In contrast, extended nucleotide context downstream of the TC motif showed strong genome-wide predictive value, with a marked preference for guanine-rich trinucleotide immediately downstream. Analysis of the upstream nucleotide context was consistent with the mutational specificity of APOBEC3A. Together, these results identify distinct structural and sequence determinants of the APOBEC-associated mutagenesis in MPXV and show that local hairpin/cruciform architecture and extended nucleotide context contribute at different genomic scales.

microbiology↗

bioRxiv2025

Bento: Benchmarking Classical and AI Docking on Drug Design-Relevant Data

AO_SCPLOWBSTRACTC_SCPLOWRecent advances in artificial intelligence have introduced deep learning and co-folding approaches for predicting protein-ligand complexes, raising the question of their applicability and how they compare with classical docking methods. In this work, we present a thorough benchmarking study of eleven tools for protein-ligand interaction prediction, spanning classical molecular docking methods, deep learning-based models, and co-folding algorithms. While most related benchmarking efforts primarily assess the generalization capacity, we extend the analysis to also evaluate the performance on drug design-relevant data and across different classes of protein-ligand complexes. Here, we introduce BO_SCPLOWENTOC_SCPLOW, a comprehensive benchmark that evaluates 11 tools for protein-ligand interaction prediction - both established and recently developed - across four test datasets and multiple derived subsets in a pocket-aware setup. We show that 1) careful dataset curation is essential - filtering by pocket structural similarity and controlling ligand complexity exposes generalization failures that are obscured in conventional benchmarks; 2) classical and deep learning-based docking tools perform similarly well on drug-like ligands, making them comparably useful for virtual screening, with physics-based methods offering a clear advantage in speed; 3) co-folding tools outperform other approaches on structurally complex ligands, whereas most methods achieve similar accuracy on regular small molecules; and 4) all methods struggle to generalize to unseen pockets, with deep learning models being the most prone to overfitting. Overall, our results show that while current docking and DL-based approaches are reliable for many drug-design-relevant scenarios, genuine pocket-level generalization remains an open challenge. BO_SCPLOWENTOC_SCPLOW provides a rigorous and transparent framework for diagnosing these limitations and guiding the development of more robust protein-ligand prediction models. The data and code of Bento are available at https://github.com/LigandPro/Bento.

bioinformatics↗

bioRxiv2025

Assessing nonlinearities in the GFP random mutagenesis landscape using the Power Transform

Epistasis, a non-additive contribution of mutations to fitness, complicates genotype-to-phenotype prediction and is often confounded by nonlinearities in phenotype measurements. The Power Transform method, particularly Box-Cox, has been used to reduce epistasis in small, combinatorially complete datasets by linearizing phenotypic scales. However, its applicability to large landscapes, especially those generated by (quasi-)random mutagenesis, remains unclear. Here, we apply both Box-Cox and Yeo-Johnson Power Transforms to the extensively characterized green fluorescent protein (GFP) fitness landscape, which contains hundreds of thousands of two- and three-dimensional combinatorially complete datasets. Surprisingly, when applied to individual hypercubes, Box-Cox reduces pairwise epistasis by 19.4% on average, whereas Yeo-Johnson - despite accommodating negative values - slightly increases epistasis (by 4.45%). Moreover, in a notable fraction of hypercubes, both methods increase the magnitude of epistatic coefficients, indicating that Power Transform does not universally reduce nonlinearity at the local scale. In contrast, applying Yeo-Johnson to the largest connected component of the GFP landscape (20,872 genotypes) successfully reduces both pairwise (by 3.86%) and third-order epistasis (by 11.6%), with the majority of coefficients decreasing in magnitude. These results demonstrate that Power Transform can be extended to large, real-world landscapes generated by random mutagenesis, but only when applied to connected sublandscapes. Our findings highlight a critical distinction between global and local linearization and caution against assuming that Power Transform always diminishes epistasis.

bioinformatics↗

bioRxiv2025

Reciprocal sign epistasis and sign epistasis may not be found in multi-peaked landscapes

Fitness landscapes provide a rigorous mathematical framework for analyzing evolutionary dynamics, including the study of epistasis, the main obstacle to predicting phenotype from genotype. In 2011, Poelwijk et al. formulated a foundational theorem stating that in any multi-peaked fitness landscape, "at least two mutations exhibit reciprocal sign epistasis" (Poelwijk et al., J. Theor. Biol., 272:141). The proof relied on the implicit assumption that neutral mutations are absent, commonly accepted in theoretical studies in evolutionary biology. In this study, we extend Poelwijk et al.s analysis by incorporating genotypes with equal fitness, specifically, accounting for neutral mutations. We demonstrate that when neutral mutations are considered, conventional pairwise reciprocal sign epistasis (RSE) may be entirely absent from a multi-peaked landscape. Instead, RSE is guaranteed only when considering "distant" RSE defined through composite mutations, wherein groups of mutations are treated collectively across all their possible combinations. Applying these concepts to empirical fitness landscapes faces a practical limitation: phenotypic measurements contain experimental noise, making mutational effects statistically indistinguishable from zero. Under such conditions, statistically significant detection of RSE in multi-peaked landscapes may be impossible even when composite mutations are considered. Theoretically, our findings imply that in the presence of neutral mutations, compensatory mutations in a multi-peaked fitness landscape need not be adjacent; rather, compensation can occur following one or more neutral steps along an evolutionary path. Practically, in real-world scenarios where fitness measurements contain uncertainty, there may be a fundamental technical limitation to detecting RSE in a statistically significant manner within multi-peaked landscapes.

evolutionary biology↗

bioRxiv2025

Insights into structures of peptide aggregation nuclei from concentration dependence of lag time

Many peptides and proteins self-assemble into large fibrillar aggregates, reaching sizes of several micrometers. This process typically involves nucleation, formation of transient oligomeric species ranging from dimers to assemblies comprising hundreds of monomers. The roles of these heterogeneous oligomers in initiating fibril growth vary significantly, as only a subset converts into primary nuclei, the smallest assemblies capable of spontaneous elongation. Consequently, the initial stages of peptide aggregation remain a critical challenge in elucidating amyloid fibril formation. Here, we analyze the aggregation lag time as a function of initial peptide concentration, employing linear regression on a double logarithmic scale to derive the critical nucleus size from the slope. We compare potential shifts in kinetic regimes with critical concentrations associated with the assembly of larger oligomers. Although peptide length has no clear correlation with nucleus size, extended helical regions may promote the formation of larger nuclei, stabilizing preliminary oligomers and prolonging the lag phase. We quantified the contributions of higher-order oligomers as on- and off-pathway species across multiple peptides, identifying a common critical micelle concentration range. We prove that linear growth models can not capture the weak concentration dependence of lag times for several peptides. We demonstrate that the inclusion of capping and fragmentation mechanisms substantially improves the plausibility of the model. Knowledge of nucleus size facilitates molecular dynamics simulations to capture transitions to fibril-prone conformations. The insights advance our understanding of amyloid nucleation, identifying toxic aggregates in human neuroglial cells, and research on drug safety and biotechnological applications.

biophysics↗

bioRxiv2025

Comment on "Inverted repeats in the monkeypox virus genome are hot spots for mutation"

Recent work proposed that APOBEC-signature mutations in monkeypox virus (MPXV) are enriched in cruci-form structures formed by inverted repeats. We independently reanalyzed the same dataset with the published settings and several complementary approaches and found no statistically supported enrichment of mutations in these structures. While a modest trend at TC motif at 3 end of hairpin loop contexts cannot be excluded, current evidence is insufficient. Additional data and careful handling of overlapping structures are needed to clarify any association.

microbiology↗

bioRxiv2025

Analysis of MPXV RNA-seq Data Reveals Lack of Evidence of APOBEC3-mediated RNA Editing

The 2022 outbreak of monkeypox virus (MPXV), a double-stranded DNA virus, is remarkable for an unusually high number of single-nucleotide substitutions compared to earlier strains, with a strong bias toward C[->]T and G[->]A transitions consistent with the APOBEC3 cytidine deaminase activity. While APOBEC3-induced mutagenesis is well documented at the DNA level, its potential impact on MPXV RNA transcripts remains unclear. To assess whether APOBEC3 enzymes act on MPXV RNA, we analyzed RNA-seq data from infected samples. The enrichment of APOBEC-signature substitutions among high-frequency mismatched positions led us to consider two possibilities: RNA editing at hotspots or fixed DNA mutations. Multiple lines of evidence support the conclusion that these substitutions arise from DNA-level mutagenesis rather than RNA editing. These include a substantial number of G[->]A substitutions remaining after normalization by gene strand direction, a largely neutral impact of substitutions on protein-coding sequences, the lack of positional correlation with transcriptional features or RNA secondary structure typically associated with APOBEC action hotspots, and an overlap with known genomic mutations in MPXV strains. Analysis of the nucleotide context of observed substitutions indicated that APOBEC3A or APOBEC3B were likely drivers of DNA-level mutagenesis. ImportanceThe 2022 monkeypox virus (MPXV) outbreak showed an unusually high number of mutations thought to result from human antiviral enzymes of the APOBEC3 family. While such mutations have been clearly documented in the viral DNA, whether APOBEC3 also edits viral messenger RNA molecules remained unclear. In this study, we analyzed multiple publicly available MPXV RNA sequencing datasets to address this question. We found that the apparent APOBEC-like changes in RNA are best explained by fixed DNA mutations rather than active RNA editing. This finding helps clarify how MPXV evolves and adapts, suggesting that APOBEC3s role in shaping the virus likely operates at the DNA level. Understanding where and how these mutations occur provides insight into the viruss interaction with the human immune system and informs future studies on viral evolution and antiviral defenses.

microbiology↗

bioRxiv2024

Convergent gene pairs restrict chromatin looping in Dictyostelium discoideum, acting as directional barriers for extrusion

Dictyostelium discoideum is a unicellular slime mold, developing into a multicellular fruiting body upon starvation. Development is accompanied by large-scale shifts in gene expression program, but underlying features of chromatin spatial organization remain unknown. Here, we report that the Dictyostelium 3D genome is organized into positionally conserved, largely consecutive, non-hierarchical and weakly insulated loops at the onset of multicellular development. The transcription level within the loop interior tends to be higher than in adjacent regions. Loop interiors frequently contain functionally linked genes and genes which coherently change expression level during development. Loop anchors are predominantly positioned by the genes in convergent orientation. Our data suggest that the loop profile may arise from the interplay between transcription and extrusion-driven chromatin folding. In particular, a convergent gene pair serves as a bidirectional extrusion barrier or a "diode" that controls passage of the cohesin extruder by relative transcription level of paired genes.

molecular biology↗

bioRxiv2023

Where to stop psi-BLAST iterations so that the found sequences remain related to the query sequence?

The BLAST program [Altschul et al., 1990, J. Mol. Biol., 215:403-410] stands as a widely used tool for the search of the most similar sequences, while the iterative {Psi}-BLAST program [Altschul et al., 1997, Nucl. Acids Res., 25:3389-3402] offers a high sensitivity for detecting remote homologs of the query sequence through an iterative usage of the BLAST search. However, the number of iterations that have to be used by the {Psi}-BLAST is rather poorly justified in the literature. Our study shows that, as the number of iterations increases, {Psi}-BLAST rapidly loses the ability to be guided by the query sequence in the search for homologs. When working with the non-redundant (nr) sequence database of 2021, {Psi}-BLAST, already after the second iteration, retains the query sequence at the top of the list of the found homologs to this sequence in only 18% of cases. Moreover, a query sequence is still listed among homologs found by {Psi}-BLAST after the recommended 10 iterations [Altschul et al., 1997, Nucl. Acids Res., 25:3389-3402] in only 42% of cases. Using a considerably smaller nr database-2011 as a reference, we reveal that these effects intensify over time. Our findings underscore the necessity for circumspection when interpreting {Psi}-BLAST outcomes; the degree of vigilance must increase with the database size. A vigilant monitoring of the position of the query sequence in the array of detected homologs is needed. We recommend using the disappearance of the query sequence from the list of homologs produced by {Psi}-BLAST as a criterion to conclude the {Psi}-BLAST iterations.

bioinformatics↗

bioRxiv2023

In silico simulations reveal molecular mechanism of uranyl ion toxicity towards DNA-binding domain of PARP-1 protein

The molecular toxicity of uranyl ion (UO22+) in living cells is mainly conditioned by its high affinity to both native and potential metal-binding sites frequently occurring in biomolecules structure. Recent advances in computational and experimental research shed light on the structural properties and functional impacts of uranyl binding to proteins, organic ligands, nucleic acids and their complexes. In the present work, we report the results of the theoretical investigation of the uranyl-mediated loss of DNA-binding activity of PARP-1, eukaryotic enzyme that participates in DNA reparation, cell differentiation, induction of inflammation, etc. Latest experimental studies showed that uranyl ion directly interacts with its DNA-binding subdomains - zinc fingers Zn1 and Zn2, - and changes their tertiary structure. Here, we propose an atomistic mechanism underlying this process and compute the free energy change along the suggested pathway to prove its relevance. According to the results of our QM/MM simulations of Zn2-UO22+ complex, uranyl ion replaces zinc in its native binding site, but the corresponding state is destroyed because of the following spontaneous internal hydrolysis of the U-Cys162 coordination bond. Although the enthalpy of hydrolysis is +2.8 kcal/mol, the final value of the free energy of the reaction constitutes -0.6 kcal/mol, due to structure loosening evidenced by solvation and configuration thermodynamic properties calculated using GIST- and MIST-based trajectory processing techniques. The subsequent reorganization of the binding site includes association of uranyl ion with the Glu190/Asp191 acidic cluster and significant perturbations in the domains tertiary structure, which further decreases the free energy of the non-functional state by 6.8 kcal/mol. The disruption of the DNA-binding interface revealed in our computational simulations is consistent with previous experimental findings and appears to be associated with the loss of the Zn2 affinity for nucleic acids.

bioinformatics↗

bioRxiv2023

The new mega dataset combined with a deep neural network makes progress in predicting the impact of single mutations on protein stability

Prediction of proteins stability change ({Delta}{Delta}G) due to single mutation is important for biotechnology, medicine, and our understanding of physics underlying protein folding. Despite the recent tremendous success in 3D protein structure prediction, the apparently simpler problem of predicting the effect of mutations on protein stability has been hampered by the low amount of experimental data. With the recent high-throughput measurements of mutational effects in mega experiment for ~850,000 mutations [Tsuboyama et al., bioRxiv, 2022] it becomes possible to apply the state-of-the-art deep learning methods. Here we explore the ability of ESM2 deep neural network architecture with added Light Attention mechanism to predict the change of protein stability due to single mutations. The resulting method ABYSSAL predicts well the data from the mega experiment (Pearson correlation 0.85) while the prediction of {Delta}{Delta}G values from previous experiments is more modest (Pearson correlation 0.50). ABYSSAL also shows a perfect satisfaction of the antisymmetry property. The ABYSSAL training demonstrated that the dataset should contain around ~100,000 data points for taking advantage of the state-of-the-art deep learning methods. Overall, our study shows great perspectives for developing the deep learning {Delta}{Delta}G predictors.

bioinformatics↗

bioRxiv2022

Composite mutations give an extra insight into epistasis

The intricate genotype-phenotype relationship has been a long-standing issue in biology, important both from the fundamental and applied points of view. One of the major irregularities hindering progress in establishing these links is epistasis - the complex and elusive interaction between mutations. Despite the vast accumulated genetic data and progress in this area, epistasis is still far from being completely understood. Epistasis can be studied quantitatively in combinatorially complete datasets, which form hypercubes in protein sequence space, where connected sequences are one mutation away from each other. However, this might be insufficient to portray the full picture of epistatic interactions. To extend the repertoire of the methods for exploring epistasis, we propose here to consider hyperrectangles, where some edges connect sequences being two or more mutations away from each other. The present work formalizes the theoretical knowledge about these novel structures and compares the amount of epistasis identified in hypercubes and hyperrectangles constructed from experimental datasets. A new algorithm, CuboidME, was developed for calculating hyperrectangles, which were then compared to hypercubes. In the experimental datasets, there were four orders of magnitude more hyperrectangles than hypercubes for the same sample size. Subsequently, we showed that for the studied datasets there is an increase in epistasis measured by epistatic coefficients in hyperrectangles compared to hypercubes. For the same datasets, hyperrectangles could find more sign epistasis than using hypercubes alone. We also show that there is a trend for increase in epistasis with increasing number of mutations being considered in a hyperrectangle. The results indicate that hyperrectangles can be used to reveal more information on epistasis in a fitness landscape, especially if it is combinatorially incomplete.

evolutionary biology↗

bioRxiv2021

Using AlphaFold to predict the impact of single mutations on protein stability and function

AlphaFold changed the field of structural biology by achieving three-dimensional (3D) structure prediction from protein sequence at experimental quality. The astounding success even led to claims that the protein folding problem is "solved". However, protein folding problem is more than just structure prediction from sequence. Presently, it is unknown if the AlphaFold-triggered revolution could help to solve other problems related to protein folding. Here we assay the ability of AlphaFold to predict the impact of single mutations on protein stability ({Delta}{Delta}G) and function. To study the question we extracted metrics from AlphaFold predictions before and after single mutation in a protein and correlated the predicted change with the experimentally known {Delta}{Delta}G values. Additionally, we correlated the AlphaFold predictions on the impact of a single mutation on structure with a large scale dataset of single mutations in GFP with the experimentally assayed levels of fluorescence. We found a very weak or no correlation between AlphaFold output metrics and change of protein stability or fluorescence. Our results imply that AlphaFold cannot be immediately applied to other problems or applications in protein folding.

bioinformatics↗

bioRxiv2021

Best templates outperform homology models in predicting the impact of mutations on protein stability

MotivationPrediction of protein stability change upon mutation ({Delta}{Delta}G) is crucial for facilitating protein engineering and understanding of protein folding principles. Robust prediction of protein folding free energy change requires the knowledge of protein three-dimensional (3D) structure. Unfortunately, protein 3D structure is not always available. In this case, one can still predict the protein stability change by constructing a homology model of the protein; however, the accuracy of homology model-based {Delta}{Delta}G predictions is unknown. The perspectives of using 3D structures of the best templates are also unclear. ResultsTo investigate these questions, we used the most popular and accurate publicly available tools: FoldX for stability change prediction and I-Tasser for homology modeling. We found that both homology models and best templates worsen the {Delta}{Delta}G prediction, with best templates performing 1.5 times better than homology models. For AlphaFold models, we also found that the best templates seem to outperform protein models. Our findings imply using the 3D structures of the best templates for {Delta}{Delta}G prediction if the 3D protein structure is unavailable. Contactd.ivankov@skoltech.ru

bioinformatics↗

bioRxiv2019

HypercubeME: two hundred million combinatorially complete datasets from a single experiment

MotivationEpistasis, the context-dependence of the contribution of an amino acid substitution to fitness, is common in evolution. To detect epistasis, fitness must be measured for at least four genotypes: the reference genotype, two different single mutants and a double mutant with both of the single mutations. For higher-order epistasis of the order n, fitness has to be measured for all 2n genotypes of an n-dimensional hypercube in genotype space forming a \"combinatorially complete dataset\". So far, only a handful of such datasets have been produced by manual curation. Concurrently, random mutagenesis experiments have produced measurements of fitness and other phenotypes in a high-throughput manner, potentially containing a number of combinatorially complete datasets.\n\nResultsWe present an effective recursive algorithm for finding all hypercube structures in random mutagenesis experimental data. To test the algorithm, we applied it to the data from a recent HIS3 protein dataset and found all 199,847,053 unique combinatorially complete genotype combinations of dimensionality ranging from two to twelve. The algorithm may be useful for researchers looking for higher-order epistasis in their high-throughput experimental data.\n\nAvailabilityhttps://github.com/ivankovlab/HypercubeME.git.

evolutionary biology↗