Biology subjects

Tosatto, S. C. E.

Publications and source records attributed to Tosatto, S. C. E..

At least 19 recordsLinked to original sources

bioRxiv2026

The Hidden Disorder Divide: Reconciling Benchmark Inconsistencies in Intrinsically Disordered Protein Binding Site Prediction

Computational predictors of protein-binding sites within intrinsically disordered regions (IDRs) show highly inconsistent performance across high-quality benchmark datasets. To understand the origins of these discrepancies, we systematically compared predictors across three independent test sets: two CAID datasets updated with the latest DisProt annotations and a composite dataset (DBs) assembled from DIBS, FuzDB, IDEAL, and MFIB. Predictors trained predominantly on DisProt data achieved substantially higher AUCs on the CAID sets but performed poorly on the DBs. In contrast, predictors trained on older, low-quality PDB-based datasets showed balanced performance across all sets, with a slight preference for DBs. Predictors with mixed training exposure displayed intermediate behavior. Through controlled experiments using identical CNN architectures and feature analysis, we demonstrate that the dominant factor driving these performance differences is the intrinsic disorder propensity of the binding sites themselves. Binding residues in DisProt-based datasets exhibit markedly higher average disorder propensity scores than those in PDB-derived datasets. This previously unrecognized selection bias -- literature studies preferentially characterizing more disordered binding sites, while PDB-derived annotations capture less disordered ones -- effectively splits IDR-protein binding sites into two distinct categories. Predictors optimized on one category therefore generalize poorly to the other. Binding-site length and sequence conservation play only minor or negligible roles in explaining the observed inconsistencies. These findings highlight a critical limitation in current benchmarking practices and training strategies for IDR-binding site prediction, underscoring the need for more balanced and disorder-aware reference datasets. Finally, the diagnostic techniques introduced here could prove valuable beyond the specific application examined in this study. The data and code used to generate all figures and tables are available at: https://github.com/NawarMalhis/HDD

bioinformatics↗

bioRxiv2026

DOME Copilot: Making transparency and reproducibility for artificial intelligence methods simple

Artificial intelligence (AI) methods are transforming life science research and witnessing unprecedented adoption across literature. While AI is driving impactful discoveries, this rapid growth in application has inundated researchers with poorly described models and datasets which lack transparency, impeding reusability and eroding trust. The DOME Recommendations aimed to address this issue and established structured reporting guidelines to standardize AI methodology descriptions. However, the need to manually create these comprehensive disclosures imposed a major bottleneck, requiring substantial effort from authors to comply. To bridge this gap, we present DOME Copilot, an open-source, easily deployable system that automates the generation of transparent AI methodology reports supplementing publications. Evaluated against a benchmark of human-created annotations, DOME Copilot was determined to match or exceed manual annotation quality across the majority of reporting fields while reducing the creation time from hours to minutes. Ultimately, the system provides a novel and scalable solution to restore confidence in complex life science AI method publications.

bioinformatics↗

bioRxiv2025

AlphaFlex: Ensembles of the human proteome representing disordered regions

More than two thirds of proteins in the human proteome are predicted to contain intrinsically disordered regions (IDRs), which lack stable folded structure. IDRs are critical for biological regulation and organization, as targets for post-translational modifications, and as mediators of biomolecular condensates. To address the pressing need for better structural models enabling functional insight, we developed AlphaFlex to model fully atomistic conformer ensembles for proteins predicted to have IDRs, modeled in the context of AlphaFold folded domains and an implicit bilayer for transmembrane proteins. The AlphaFlex resource provides conformational ensembles of human proteins from the AlphaFold database with identified IDRs in the Protein Ensemble Database that is mirrored in UniProt. This transformative resource of AlphaFlex ensembles provides physically and biologically relevant full-length models for IDR proteins, including scaffold proteins, those with IDR:folded-domain interactions, regulatory and condensate proteins requiring exposed binding elements, conditionally folding IDRs, and transmembrane proteins containing IDRs.

biochemistry↗

bioRxiv2025

GeomeTRe: accurate calculation of geometrical descriptors of tandem repeat proteins

Structured tandem repeat proteins (STRPs) are characterized by preserved structural motifs arranged in a modular way. The structural and functional diversity of STRPs makes them particularly important for studying evolution, novel structure and function relationships, and ultimately for designing new synthetic proteins with specific functions. One crucial aspect of their classification is the estimation of geometrical parameters, which can provide better insight into their properties and the relationship between the spatial arrangement of repeated units and protein function. Calculating geometric descriptors for STRPs is challenging because naturally occurring repeats are not perfect and often contain insertions and deletions. Existing tools for predicting structural symmetry work well on simple cases but often fail for most natural proteins. Here we present GeomeTRe, an algorithm that calculates geometrical descriptors such as curvature (yaw), twist (roll), and pitch for a protein structure with known repeat unit positions. The algorithm simulates the movement of consecutive units, identifies rotational axes, and calculates the corresponding Tait-Bryan angles. The parameters of GeomeTRe can enhance STRP annotation and classification by identifying variations in geometric arrangements among different functional groups. The package is fast and suitable for processing large protein structure datasets when repeat region information (e.g., from RepeatsDB) is available. GeomeTRe is available as a Python package; source code and documentation can be found at https://github.com/BioComputingUP/GeomeTRe

bioinformatics↗

bioRxiv2024

STRPsearch: fast detection of structured tandem repeat proteins

MotivationState-of-the-art prediction methods are generating millions of publicly available protein structures. Structured Tandem Repeats Proteins (STRPs) constitute a subclass of tandem repeats characterized by repetitive structural motifs. STRPs exhibit distinct propensities for secondary structure and form regular tertiary structures, often comprising large molecular assemblies. They can perform important and diverse biological functions due to their highly degenerated sequences, which maintain a similar structure while displaying a variable number of repeat units. This suggests a disconnection between structural size and protein function. However, automatic detection of STRPs remains challenging with current state-of-the-art tools due to their lack of accuracy and long execution times, hindering their application on large datasets. In most cases, manual curation is the most accurate method for detecting and classifying them, making it impossible to inspect millions of structures. ResultsWe present STRPsearch, a novel computational tool for rapid identification, classification, and mapping of STRPs. Leveraging the manually curated entries in RepeatsDB as the known conformational space of the STRPs, STRPsearch utilizes the latest advancements in structural alignment techniques for a fast and accurate detection of repeated structural motifs in protein structures, followed by an innovative approach to map units and insertions through the generation of TM-score graphs. STRPsearch can serve researchers in structural bioinformatics and protein science as an efficient and practical tool for analysis and detection of STRPs. Availability and implementationSTRPsearch is coded in Python, all the scripts and the associated documentation are available at https://github.com/BioComputingUP/STRPsearch. Contactalexander.monzon@unipd.it, silvio.tosatto@unipd.it Supplementary informationSupplementary data are available..

bioinformatics↗

bioRxiv2023

Patient-specific simulation of Retinal Hemangioblastoma provides new perspectives on the role ofantiangiogenic therapy

Retinal Hemangioblastoma (RH) is the most frequent manifestation of the von Hippel-Lindau syndrome (VHL), a rare disease associated with the germline mutation of the von Hippel-Lindau protein (pVHL). An emblematic feature of RH is the high vascularity, which is easily explained by the overexpression of angiogenic factors (AFs) arising from the pVHL impairment. The introduction of Optical Coherence Tomography Angiography (OCTA) allowed observing this feature with exceptional detail. However, our understanding of RH is limited by the absence of an animal model fully recapitulating the tumor. Here, we exploit a cancer mathematical model as an alternative way to explore RH development and angiogenesis. We derived our model from the agreed pathology for this tumor and compared our results with patient-specific OCTA images. Our simulations closely resemble the medical images, proving the capability of our model to recapitulate RH pathology. Our results also suggest that angiogenesis in RH occurs suddenly when the tumor reaches a critical mass, with full capillary invasion in the order of days. These findings open a new perspective on the critical role of time in antiangiogenic therapy in RH, which has resulted ineffective. Indeed, it might be that when RH is diagnosed, angiogenesis is already too advanced to be effectively targeted with this mean.

cancer biology↗

bioRxiv2022

Pipeline for transferring annotations between proteins beyond globular domains

BackgroundDisProt is the primary repository of Intrinsically Disordered Proteins. This database is manually curated and the annotations there have strong experimental support. Currently DisProt contains a relatively small number of proteins highlighting the importance of transferring verified disorder and other annotations, in such a way as to increase the number of proteins that could benefit from this valuable information. While the principles and practicalities of homology transfer are well-established for globular proteins, these are largely lacking for disordered proteins. MethodsWe used DisProt to evaluate the transferability of the annotation terms to orthologous proteins. For each protein, we looked for their orthologs, with the assumption that they will have a similar function. Then, for each protein and their orthologs we made multiple sequence alignments (MSAs). Global and regional quality of the MSAs was evaluated with the NorMD score. ResultsWe have designed a pipeline to obtain good quality MSAs and to transfer annotations from any protein to their orthologs. Applying the pipeline to DisProt proteins, from the 1931 entries with 5,623 annotations we can reach 97,555 orthologs and transfer a total of 301,190 terms by homology. We also provide a web server for consulting the results of DisProt proteins and execute the pipeline for any other protein. The server Homology Transfer IDP (HoTIDP) is accessible at http://hotidp.leloir.org.ar.

bioinformatics↗

bioRxiv2022

PDB ProtVista: A reusable and open-source sequence feature viewer

The PDB ProtVista is a reusable and customisable sequence feature viewer that provides intuitive and detailed 2D visualisation of residue-level annotations while supporting interactive communication with 3D viewers. The Protein Data Bank in Europe (PDBe) team develops and maintains PDB ProtVista. Several public web services use it to display structural and functional annotations such as macromolecular interaction interfaces, intrinsic disorder predictions, sequence variants, and sequence conservation. The PDB ProtVista is freely available from https://github.com/PDBeurope/protvista-pdb. We provide extensive documentation and step-by-step user guides on integrating PDB ProtVista with existing web applications and 3D molecular viewers. We also offer examples of displaying the users custom data and functional annotations for PDB and UniProt entries powered by a rich set of PDBe API endpoints.

bioinformatics↗

bioRxiv2022

Intrinsic Protein Disorder, Conditional Folding and AlphaFold2

Intrinsically disordered regions (IDRs) defying the traditional protein structure-function paradigm have been difficult to analyze. AlphaFolds recent breakthrough in predicting protein structures accurately offers a fresh perspective on IDR prediction as assessed on the CAID dataset. Surprisingly, AlphaFold is highly competitive for predicting both IDRs and conditionally folded regions, demonstrating the plasticity of the disorder to structure continuum.

bioinformatics↗

bioRxiv2021

APICURON: a database to credit and acknowledge the work of biocurators

APICURON is an open and freely accessible resource that tracks and credits the work of biocurators across multiple participating knowledgebases. Biocuration is essential to extract knowledge from research data and make it available in a structured and standardized way to the scientific community. However, processing biological data - mainly from literature - requires a huge effort which is difficult to quantify and acknowledge. APICURON collects biocuration events from third party resources and aggregates this information, spotlighting biocurator contributions. APICURON promotes biocurator engagement implementing gamification concepts like badges, medals and leaderboards and at the same time provides a monitoring service for registered resources and for biocurators themselves. APICURON adopts a data model that is flexible enough to represent and track the majority of biocuration activities. Biocurators are identified through their ORCID. The definition of curation events, scoring systems and rules for assigning badges and medals are resource-specific and easily customizable. Registered resources can transfer curation activities on the fly through a secure and robust Application Programming Interface (API). Here we show how simple and effective it is to connect a resource to APICURON describing the DisProt database of intrinsically disordered proteins as a use case. We believe APICURON will provide biological knowledgebases with a service to recognize and credit the effort of their biocurators, monitor their activity and promote curators engagement. Database URL: https://apicuron.org

bioinformatics↗

bioRxiv2020

HIF1a-dependent induction of the mitochondrial chaperone TRAP1 regulates bioenergetic adaptations to hypoxia

The mitochondrial paralog of the Hsp90 chaperone family TRAP1 is often induced in tumors, but the mechanisms controlling its expression, as well as its physiological functions remain poorly understood. Here, we find that TRAP1 is highly expressed in the early stages of Zebrafish development, and its ablation delays embryogenesis while increasing mitochondrial respiration of fish larvae. TRAP1 expression is enhanced by hypoxic conditions both in developing embryos and in cancer models of Zebrafish and mammals. The TRAP1 promoter contains evolutionary conserved hypoxic responsive elements, and HIF1 stabilization increases TRAP1 levels. TRAP1 inhibition by selective compounds or by genetic knock-out maintains a high level of respiration in Zebrafish embryos after exposure to hypoxia. Our data identify TRAP1 as a primary regulator of mitochondrial bioenergetics in highly proliferating cells following reductions in oxygen tension and HIF1 stabilization.

biochemistry↗

bioRxiv2020

Critical Assessment of Protein Intrinsic Disorder Prediction

Intrinsically disordered proteins defying the traditional protein structure-function paradigm represent a challenge to study experimentally. As a large part of our knowledge rests on computational predictions, it is crucial for their accuracy to be high. The Critical Assessment of protein Intrinsic Disorder prediction (CAID) experiment was established as a community-based blind test to determine the state of the art in predicting intrinsically disordered regions in proteins and the subset of disordered residues involved in binding other molecules. A total of 43 methods, 32 for disorder and 11 for binding regions, were evaluated on a dataset of 646 novel manually curated proteins from DisProt. The best methods use deep learning techniques and significantly outperform widely used earlier physicochemical methods across different types of targets. Disordered binding regions remain hard to predict correctly. Depending on the definition used, the top disorder predictor has an FMax of 0.483 (DisProt) or 0.792 (DisProt-PDB). As the top binding predictor only attains an FMax of 0.231, this suggests significant potential for improvement. Intriguingly, computing times among the top performing methods vary by up to four orders of magnitude.

bioinformatics↗

bioRxiv2020

Intrinsically disordered protein ensembles shape evolutionary rates revealing conformational patterns

Intrinsically disordered proteins (IDPs) lack stable tertiary structure under physiological conditions. The unique composition and complex dynamical behaviour of IDPs make them a challenge for structural biology and molecular evolution studies. Using NMR ensembles, we found that IDPs evolve under a strong site-specific evolutionary rate heterogeneity, mainly originated by different constraints derived from their inter-residue contacts. Evolutionary rate profiles correlate with the experimentally observed conformational diversity of the protein, allowing the description of different conformational patterns possibly related to their structure-function relationships. The correlation between evolutionary rates and contact information improves when structural information is taken not from any individual conformer or the whole ensemble, but from combining a limited number of conformers. Our results suggest that residue contacts in disordered regions constrain evolutionary rates to conserve the dynamic behaviour of the ensemble and that evolutionary rates can be used as a proxy for the conformational diversity of IDPs. Significance StatementIntrinsically disordered proteins (IDPs) challenge the structure-function relationship paradigm. In this work we found that individual sites of IDPs evolve under a strong rate heterogeneity, mainly due to the structural constraints imposed by contacts between their residues. This can be better explained if the contacts are taken from selected subsets of their alternative native conformations, rather than from individual conformations or the whole native ensemble. From an evolutionary point of view, this result indicates that experimentally-based ensembles are redundant. We also observed that the evolutionary rates follow the structural variability between conformers, unveiling conformational preferences. Our results set the stage for establishing novel evolutionary-based methods to study IDP ensembles.

bioinformatics↗

bioRxiv2020

The E3 ubiquitin-protein ligase MDM2 is a novel interactor of the von Hippel-Lindau tumor suppressor

Mutations of the von Hippel-Lindau (pVHL) tumor suppressor are causative of a familiar predisposition to develop different types of cancer. pVHL is mainly known for its role in regulating hypoxia-inducible factor 1- (HIF-1) degradation, thus modulating the hypoxia response. There are different pVHL isoforms, including pVHL30 and pVHL19. However, little is known about isoform-specific functions and protein-protein interactions. Integrating in silico predictions with in vitro and in vivo assays, we describe a novel interaction between pVHL and mouse double minute 2 homolog (MDM2). Importantly, we found that pVHL30, and not pVHL19, forms a complex with MDM2, and that the N-terminal acidic tail of pVHL30 is required for its association with MDM2. Further, we demonstrate that an intrinsically disordered region upstream of the tetramerization domain of MDM2 is responsible for its isoform-specific association with pVHL30. This region is highly conserved in higher mammals, including primates, similarly to what has been already proposed for the N-terminal tail of pVHL30. Finally, we show that overexpression of pVHL30 and MDM2 together reduces cell proliferation, suggesting a synergistic effect of these E3 ubiquitin ligases. Collectively, our data support the idea that pVHL30 plays a role in MDM2 regulation, suggesting a wider interplay among hypoxia sensing and cell cycle regulation.

biochemistry↗

bioRxiv2020

Experimentally determined long intrinsically disordered protein regions are now abundant in the Protein Data Bank

Intrinsically disordered protein regions are commonly defined from missing electron density in X-ray structures. Experimental evidence for long disorder regions (LDRs) of at least 30 residues was so far limited to a less than a thousand manually curated proteins. Here, we describe a comprehensive and large-scale analysis of experimental LDRs for 3,133 unique proteins, demonstrating an increasing coverage of intrinsic disorder in the Protein Data Bank (PDB) in the last decade. The results suggest that long missing residue regions are a good quality source to annotate intrinsically disordered regions and perform functional analysis in large data sets. The consensus approach used to define LDRs allows to evaluate context dependent disorder and provide a common definition at the protein level.

bioinformatics↗

bioRxiv2020

Assessing predictors for new post translational modification sites: a case study on hydroxylation

Post-translational modification (PTM) sites have become popular for predictor development. However, with the exception of phosphorylation and a handful of other examples, PTMs suffer from a limited number of available training examples and their sparsity in protein sequences. Here, proline hydroxylation is taken as an example to compare different methods and evaluate their performance on new experimentally determined sites. As a proxy for an effective experimental design, predictors require both high specificity and sensitivity. However, the self-reported performance is often not indicative of prediction quality and detection of new sites is not guaranteed. We have benchmarked seven published hydroxylation site predictors on two newly constructed independent datasets. The self-reported performance widely overestimates the real accuracy measured on independent datasets. No predictor performs better than random on new examples, indicating the refined models are not sufficiently general to detect new sites. The number of false positives is high and precision low, in particular for non-collagen proteins whose motifs are not conserved. In short, existing predictors for hydroxylation sites do not appear to generalize to new data. Caution is advised when dealing with PTM predictors in the absence of independent evaluations, in particular for unique specific sites such as those involved in signalling. Author SummaryMachine learning methods are extensively used by biologists to design and interpret experiments. Predictors which take the only sequence as input are of particular interest due to the large amount of sequence data available and self-reported performance is often very high. In this work, we evaluated post-translational modification (PTM) predictors for hydroxylation sites and found that they perform no better than random, in strong contrast to performances reported in the original publications. PTMs are chemical amino acids alterations providing the cell with conditional mechanisms to fine tune protein function, thereby regulating complex biological processes such as signalling and cell cycle. Hydroxylation sites are a good PTM test case due to the availability of a range of predictors and an abundance of newly experimentally detected modification sites. Poor performances in our results highlight the overlooked problem of predicting PTMs when best practices are not followed and training data are likely incomplete. Experimentalists should be careful when using PTM predictors blindly and more independent assessments are needed to separate the wheat from the chaff in the field.

bioinformatics↗

bioRxiv2019

Characterization of Intellectual disability and Autism comorbidity through gene panel sequencing

Intellectual disability (ID) and autism spectrum disorder (ASD) are clinically and genetically heterogeneous diseases. Recent whole exome sequencing studies indicated that genes associated with different neurological diseases are shared across disorders and converge on common functional pathways. Using the Ion Torrent platform, we developed a low-cost next generation sequencing (NGS) gene panel that has been transferred into clinical practice, replacing single disease gene analyses for the early diagnosis of individuals with ID/ASD. The gene panel was designed using an innovative in silico approach based on disease networks and mining data from public resources to score disease-gene associations. We analyzed 150 unrelated individuals with ID and/or ASD and a confident diagnosis has been reached in 26 cases (17%). Likely pathogenic mutations have been identified in another 15 patients, reaching a total diagnostic yield of 27%. Our data also support the pathogenic role of genes recently proposed to be involved in ASD. Although many of the identified variants need further investigation to be considered disease-causing, our results indicate the efficiency of the targeted gene panel on the identification of novel and rare variants in patients with ID and ASD.

genetics↗

bioRxiv2018

In silico prediction of blood cholesterol levels from genotype data

In this work we present a framework for blood cholesterol levels prediction from genotype data. The predictor is based on an algorithm for cholesterol metabolism simulation available in literature, implemented and optimized by our group in R language. Main weakness of the former simulation algorithm was the need of experimental data to simulate mutations in genes altering the cholesterol metabolism. This caveat strongly limited the application of the model in the clinical practice. In this work we present how this limitation could be bypassed thanks to an optimization of model parameters based on patients cholesterol levels retrieved from literature. Prediction performance has been assessed taking in consideration several scoring indices currently used for performance evaluation of machine learning methods. Our assessment shows how the optimization phase improved model performance, compared to the original version available in literature.

systems biology↗