MOTIVATION: Rare diseases collectively affect 5% of the population. However, fewer than 50% of rare disease patients receive a molecular diagnosis after whole genome sequencing. Supervised machine learning is a valuable approach for the pathogenicity scoring of human genetic variants. However, existing methods are often trained on curated but limited central repositories, resulting in poor accuracy when tested on external cohorts. Yet, large collections of variants generated at hospitals and research institutions remain inaccessible to machine-learning purposes because of privacy and legal constraints. Federated learning (FL) algorithms have been recently developed enabling institutions to collaboratively train models without sharing their local datasets. RESULTS: Here, we present a proof-of-concept study evaluating the effectiveness of FL for the clinical classification of genetic variants. A comprehensive array of diverse FL strategies was assessed for coding and non-coding Single Nucleotide Variants as well as Copy Number Variants. Our results showed that federated models generally achieved comparable or superior performance to traditional centralized learning. In addition, federated models reached a robust generalization to independent sets with smaller data fractions as compared to their centralized model counterparts. Our findings support the adoption of FL to establish secure multi-institutional collaborations in human variant interpretation. AVAILABILITY AND IMPLEMENTATION: All source code required to reproduce the results presented in this article, implemented in Python, is available under the GNU General Public License v3 at https://github.com/RausellLab/FedLearnVar.

Predicting the difference in thermodynamic stability between protein variants is crucial for protein design and understanding the genotype-phenotype relationships. So far, several computational tools have been created to address this task. Nevertheless, most of them have been trained or optimized on the same and 'all' available data, making a fair comparison unfeasible. Here, we introduce a novel dataset, collected and manually cleaned from the latest version of the ThermoMutDB database, consisting of 669 variants not included in the most widely used training datasets. The prediction performance and the ability to satisfy the antisymmetry property by considering both direct and reverse variants were evaluated across 21 different tools. The Pearson correlations of the tested tools were in the ranges of 0.21-0.5 and 0-0.45 for the direct and reverse variants, respectively. When both direct and reverse variants are considered, the antisymmetric methods perform better achieving a Pearson correlation in the range of 0.51-0.62. The tested methods seem relatively insensitive to the physiological conditions, performing well also on the variants measured with more extreme pH and temperature values. A common issue with all the tested methods is the compression of the $\Delta \Delta G$ predictions toward zero. Furthermore, the thermodynamic stability of the most significantly stabilizing variants was found to be more challenging to predict. This study is the most extensive comparisons of prediction methods using an entirely novel set of variants never tested before.

SUMMARY: Identifying pathogenic variants and annotating them is a major challenge in human genetics, especially for the non-coding ones. Several tools have been developed and used to predict the functional effect of genetic variants. However, the calibration assessment of the predictions has received little attention. Calibration refers to the idea that if a model predicts a group of variants to be pathogenic with a probability P, it is expected that the same fraction P of true positive is found in the observed set. For instance, a well-calibrated classifier should label the variants such that among the ones to which it gave a probability value close to 0.7, approximately 70% actually belong to the pathogenic class. Poorly calibrated algorithms can be misleading and potentially harmful for clinical decision making. AVALIABILITY AND IMPLEMENTATION: The dataset used for testing the methods is available through the DOI:10.5281/zenodo.4448197. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The accurate characterization of the translational mechanism is crucial for enhancing our understanding of the relationship between genotype and phenotype. In particular, predicting the impact of the genetic variants on gene expression will allow to optimize specific pathways and functions for engineering new biological systems. In this context, the development of accurate methods for predicting the translation efficiency and/or protein expression from the nucleotide sequence is a key challenge in computational biology. In this work we present PGExpress, a new regression method for predicting the log2-fold-change of the translation efficiency of an mRNA sequence in E. coli. PGExpress algorithm takes as input 12 features corresponding to the predicted RNA secondary structure and anti-Shine-Dalgarno hybridization free energies. The method was trained on a set of 1,772 sequence variants (WT-High)of 137 essential E. coli genes. For each gene, we considered 13 sequence variants of the first 33 nucleotides encoding for the same amino acids followed by the superfolder GFP. Each gene variant is represented sequence blocks that include the Ribosome Binding Site (RBS), the first 33 nucleotides of the coding region (C33), the remaining part of the coding region (CC), and their combinations. Our gradient-boosting-based tool (PGExpress) was trained using a 10-fold gene-based cross-validation procedure on the WT-High dataset. In this test PGExpress achieved a correlation coefficient of 0.60, with a Root Mean Square Error (RMSE)of 1.3. When the regression task is cast as a classification problem, PGExpress reached an overall accuracy of 0.74 a Matthews correlation coefficient 0.48 and an Area Under the Receiver Operating Characteristic Curve (AUC)of 0.81. In the regression task, PGExpress results in better performance than RBSCalculator in the prediction of the log2-fold-change of the translational efficiency and its variation on the WT-High dataset. Finally, we validated our method by performing in-house experiments on five newly generated mRNA sequence variants. The predictions of the expression level of the new variants are in agreement with our experimental results in E. coli.

BACKGROUND: Predicting the effect of single point variations on protein stability constitutes a crucial step toward understanding the relationship between protein structure and function. To this end, several methods have been developed to predict changes in the Gibbs free energy of unfolding (∆∆G) between wild type and variant proteins, using sequence and structure information. Most of the available methods however do not exhibit the anti-symmetric prediction property, which guarantees that the predicted ∆∆G value for a variation is the exact opposite of that predicted for the reverse variation, i.e., ∆∆G(A → B) = -∆∆G(B → A), where A and B are amino acids. RESULTS: Here we introduce simple anti-symmetric features, based on evolutionary information, which are combined to define an untrained method, DDGun (DDG untrained). DDGun is a simple approach based on evolutionary information that predicts the ∆∆G for single and multiple variations from sequence and structure information (DDGun3D). Our method achieves remarkable performance without any training on the experimental datasets, reaching Pearson correlation coefficients between predicted and measured ∆∆G values of ~ 0.5 and ~ 0.4 for single and multiple site variations, respectively. Surprisingly, DDGun performances are comparable with those of state of the art methods. DDGun also naturally predicts multiple site variations, thereby defining a benchmark method for both single site and multiple site predictors. DDGun is anti-symmetric by construction predicting the value of the ∆∆G of a reciprocal variation as almost equal (depending on the sequence profile) to -∆∆G of the direct variation. This is a valuable property that is missing in the majority of the methods. CONCLUSIONS: Evolutionary information alone combined in an untrained method can achieve remarkably high performances in the prediction of ∆∆G upon protein mutation. Non-trained approaches like DDGun represent a valid benchmark both for scoring the predictive power of the individual features and for assessing the learning capability of supervised methods.

Abstract Summary: Accurate prediction of amyloid-forming amino acid sequences remains an important challenge. We here present an online database that provides open access to the largest set of experimentally characterized amyloid forming hexapeptides. To this end, we expanded our previous set of 280 hexapeptides used to develop the Waltz algorithm with 89 peptides from literature review and by systematic experimental characterisation of the aggregation of 720 hexapeptides by transmission electron microscopy, dye binding and Fourier transform infrared spectroscopy. This brings the total number of experimentally characterized hexapeptides in the WALTZ-DB database to 1089, of which 244 are annotated as positive for amyloid formation. Availability and implementation: The WALTZ-DB database is freely available without any registration requirement at http://waltzdb.switchlab.org. Contact: [email protected] or [email protected]

Abstract Contact: [email protected] or [email protected] .

MOTIVATION: The recent advance in high-throughput sequencing technologies is generating a huge amount of data that are becoming an important resource for deciphering the genotype underlying a given phenotype. Genome sequencing has been extensively applied to the study of the cancer genomes. Although a few methods have been already proposed for the detection of cancer-related genes, their automatic identification is still a challenging task. Using the genomic data made available by The Cancer Genome Atlas Consortium (TCGA), we propose a new prioritization approach based on the analysis of the distribution of putative deleterious variants in a large cohort of cancer samples. RESULTS: In this paper, we present ContastRank, a new method for the prioritization of putative impaired genes in cancer. The method is based on the comparison of the putative defective rate of each gene in tumor versus normal and 1000 genome samples. We show that the method is able to provide a ranked list of putative impaired genes for colon, lung and prostate adenocarcinomas. The list significantly overlaps with the list of known cancer driver genes previously published. More importantly, by using our scoring approach, we can successfully discriminate between TCGA normal and tumor samples. A binary classifier based on ContrastRank score reaches an overall accuracy >90% and the area under the curve (AUC) of receiver operating characteristics (ROC) >0.95 for all the three types of adenocarcinoma analyzed in this paper. In addition, using ContrastRank score, we are able to discriminate the three tumor types with a minimum overall accuracy of 77% and AUC of 0.83. CONCLUSIONS: We describe ContrastRank, a method for prioritizing putative impaired genes in cancer. The method is based on the comparison of exome sequencing data from different cohorts and can detect putative cancer driver genes. ContrastRank can also be used to estimate a global score for an individual genome about the risk of adenocarcinoma based on the genetic variants information from a whole-exome VCF (Variant Calling Format) file. We believe that the application of ContrastRank can be an important step in genomic medicine to enable genome-based diagnosis. AVAILABILITY AND IMPLEMENTATION: The lists of ContrastRank scores of all genes in each tumor type are available as supplementary materials. A webserver for evaluating the risk of the three studied adenocarcinomas starting from whole-exome VCF file is under development. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Aligning RNAs is useful to search for homologous genes, study evolutionary relationships, detect conserved regions and identify any patterns that may be of biological relevance. Poor levels of conservation among homologs, however, make it difficult to compare RNA sequences, even when considering closely evolutionary related sequences. RESULTS: We describe SARA-Coffee, a tertiary structure-based multiple RNA aligner, which has been validated using BRAliDARTS, a new benchmark framework designed for evaluating tertiary structure-based multiple RNA aligners. We provide two methods to measure the capacity of alignments to match corresponding secondary and tertiary structure features. On this benchmark, SARA-Coffee outperforms both regular aligners and those using secondary structure information. Furthermore, we show that on sequences in which <60% of the nucleotides form base pairs, primary sequence methods usually perform better than secondary-structure aware aligners. AVAILABILITY AND IMPLEMENTATION: The package and the datasets are available from http://www.tcoffee.org/Projects/saracoffee and http://structure.biofold.org/sara/.

SUMMARY: The understanding of the biological role of RNA molecules has changed. Although it is widely accepted that RNAs play important regulatory roles without necessarily coding for proteins, the functions of many of these non-coding RNAs are unknown. Thus, determining or modeling the 3D structure of RNA molecules as well as assessing their accuracy and stability has become of great importance for characterizing their functional activity. Here, we introduce a new web application, WebRASP, that uses knowledge-based potentials for scoring RNA structures based on distance-dependent pairwise atomic interactions. This web server allows the users to upload a structure in PDB format, select several options to visualize the structure and calculate the energy profile. The server contains online help, tutorials and links to other related resources. We believe this server will be a useful tool for predicting and assessing the quality of RNA 3D structures. AVAILABILITY AND IMPLEMENTATION: The web server is available at http://melolab.org/webrasp. It has been tested on the most popular web browsers and requires Java plugin for Jmol visualization.

An international consortium released the first draft sequence of the human genome 10 years ago. Although the analysis of this data has suggested the genetic underpinnings of many diseases, we have not yet been able to fully quantify the relationship between genotype and phenotype. Thus, a major current effort of the scientific community focuses on evaluating individual predispositions to specific phenotypic traits given their genetic backgrounds. Many resources aim to identify and annotate the specific genes responsible for the observed phenotypes. Some of these use intra-species genetic variability as a means for better understanding this relationship. In addition, several online resources are now dedicated to collecting single nucleotide variants and other types of variants, and annotating their functional effects and associations with phenotypic traits. This information has enabled researchers to develop bioinformatics tools to analyze the rapidly increasing amount of newly extracted variation data and to predict the effect of uncharacterized variants. In this work, we review the most important developments in the field--the databases and bioinformatics tools that will be of utmost importance in our concerted effort to interpret the human variome.

MOTIVATION: Over the recent years, the vision that RNA simply serves as information transfer molecule has dramatically changed. The study of the sequence/structure/function relationships in RNA is becoming more important. As a direct consequence, the total number of experimentally solved RNA structures has dramatically increased and new computer tools for predicting RNA structure from sequence are rapidly emerging. Therefore, new and accurate methods for assessing the accuracy of RNA structure models are clearly needed. RESULTS: Here, we introduce an all-atom knowledge-based potential for the assessment of RNA three-dimensional (3D) structures. We have benchmarked our new potential, called Ribonucleic Acids Statistical Potential (RASP), with two different decoy datasets composed of near-native RNA structures. In one of the benchmark sets, RASP was able to rank the closest model to the X-ray structure as the best and within the top 10 models for ∼93 and ∼95% of decoys, respectively. The average correlation coefficient between model accuracy, calculated as the root mean square deviation and global distance test-total score (GDT-TS) measures of C3' atoms, and the RASP score was 0.85 and 0.89, respectively. Based on a recently released benchmark dataset that contains hundreds of 3D models for 32 RNA motifs with non-canonical base pairs, RASP scoring function compared favorably to ROSETTA FARFAR force field in the selection of accurate models. Finally, using the self-splicing group I intron and the stem-loop IIIc from hepatitis C virus internal ribosome entry site as test cases, we show that RASP is able to discriminate between known structure-destabilizing mutations and compensatory mutations. AVAILABILITY: RASP can be readily applied to assess all-atom or coarse-grained RNA structures and thus should be of interest to both developers and end-users of RNA structure prediction methods. The computer software and knowledge-based potentials are freely available at http://melolab.org/supmat.html. CONTACT: [email protected]; [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Widespread availability of low-cost, full genome sequencing will introduce new challenges for bioinformatics. RESULTS: This review outlines recent developments in sequencing technologies and genome analysis methods for application in personalized medicine. New methods are needed in four areas to realize the potential of personalized medicine: (i) processing large-scale robust genomic data; (ii) interpreting the functional effect and the impact of genomic variation; (iii) integrating systems data to relate complex genetic interactions with phenotypes; and (iv) translating these discoveries into medical practice. CONTACT: [email protected]

BACKGROUND: Single Nucleotide Polymorphisms (SNPs) are an important source of human genome variability. Non-synonymous SNPs occurring in coding regions result in single amino acid polymorphisms (SAPs) that may affect protein function and lead to pathology. Several methods attempt to estimate the impact of SAPs using different sources of information. Although sequence-based predictors have shown good performance, the quality of these predictions can be further improved by introducing new features derived from three-dimensional protein structures. RESULTS: In this paper, we present a structure-based machine learning approach for predicting disease-related SAPs. We have trained a Support Vector Machine (SVM) on a set of 3,342 disease-related mutations and 1,644 neutral polymorphisms from 784 protein chains. We use SVM input features derived from the protein's sequence, structure, and function. After dataset balancing, the structure-based method (SVM-3D) reaches an overall accuracy of 85%, a correlation coefficient of 0.70, and an area under the receiving operating characteristic curve (AUC) of 0.92. When compared with a similar sequence-based predictor, SVM-3D results in an increase of the overall accuracy and AUC by 3%, and correlation coefficient by 0.06. The robustness of this improvement has been tested on different datasets and in all the cases SVM-3D performs better than previously developed methods even when compared with PolyPhen2, which explicitly considers in input protein structure information. CONCLUSION: This work demonstrates that structural information can increase the accuracy of disease-related SAPs identification. Our results also quantify the magnitude of improvement on a large dataset. This improvement is in agreement with previously observed results, where structure information enhanced the prediction of protein stability changes upon mutation. Although the structural information contained in the Protein Data Bank is limiting the application and the performance of our structure-based method, we expect that SVM-3D will result in higher accuracy when more structural date become available.

BACKGROUND: In recent years, the number of available RNA structures has rapidly grown reflecting the increased interest on RNA biology. Similarly to the studies carried out two decades ago for proteins, which gave the fundamental grounds for developing comparative protein structure prediction methods, we are now able to quantify the relationship between sequence and structure conservation in RNA. RESULTS: Here we introduce an all-against-all sequence- and three-dimensional (3D) structure-based comparison of a representative set of RNA structures, which have allowed us to quantitatively confirm that: (i) there is a measurable relationship between sequence and structure conservation that weakens for alignments resulting in below 60% sequence identity, (ii) evolution tends to conserve more RNA structure than sequence, and (iii) there is a twilight zone for RNA homology detection. DISCUSSION: The computational analysis here presented quantitatively describes the relationship between sequence and structure for RNA molecules and defines a twilight zone region for detecting RNA homology. Our work could represent the theoretical basis and limitations for future developments in comparative RNA 3D structure prediction.

BACKGROUND: A basic question of protein structural studies is to which extent mutations affect the stability. This question may be addressed starting from sequence and/or from structure. In proteomics and genomics studies prediction of protein stability free energy change (DeltaDeltaG) upon single point mutation may also help the annotation process. The experimental DeltaDeltaG values are affected by uncertainty as measured by standard deviations. Most of the DeltaDeltaG values are nearly zero (about 32% of the DeltaDeltaG data set ranges from -0.5 to 0.5 kcal/mole) and both the value and sign of DeltaDeltaG may be either positive or negative for the same mutation blurring the relationship among mutations and expected DeltaDeltaG value. In order to overcome this problem we describe a new predictor that discriminates between 3 mutation classes: destabilizing mutations (DeltaDeltaG<-1.0 kcal/mol), stabilizing mutations (DeltaDeltaG>1.0 kcal/mole) and neutral mutations (-1.0</=DeltaDeltaG</=1.0 kcal/mole). RESULTS: In this paper a support vector machine starting from the protein sequence or structure discriminates between stabilizing, destabilizing and neutral mutations. We rank all the possible substitutions according to a three state classification system and show that the overall accuracy of our predictor is as high as 56% when performed starting from sequence information and 61% when the protein structure is available, with a mean value correlation coefficient of 0.27 and 0.35, respectively. These values are about 20 points per cent higher than those of a random predictor. CONCLUSIONS: Our method improves the quality of the prediction of the free energy change due to single point protein mutations by adopting a hypothesis of thermodynamic reversibility of the existing experimental data. By this we both recast the thermodynamic symmetry of the problem and balance the distribution of the available experimental measurements of free energy changes. This eliminates possible overestimations of the previously described methods trained on an unbalanced data set comprising a number of destabilizing mutations higher than stabilizing ones.

The detection of remote homolog pairs of proteins using computational methods is a pivotal problem in structural bioinformatics, aiming to compute protein folds on the basis of information in the database of known structures. In the last 25 years, several methods have been developed to tackle this problem, based on different approaches including sequence-sequence alignments and/or structure comparison. In this article, we will briefly discuss When, Why, Where and How (WWWH) to perform remote homology search, reviewing some of the most widely adopted computational approaches. The specific aim is highlighting the basic criteria implemented by different research groups and commenting on the status of the art as well as on still-open questions.

UNLABELLED: K-Fold is a tool for the automatic prediction of the protein folding kinetic order and rate. The tool is based on a support vector machine (SVM) that was trained on a data set of 63 proteins, whose 3D structure and folding mechanism are known from experiments already described in the literature. The method predicts whether a protein of known atomic structure folds according to a two-state or a multi-state kinetics and correctly classifies 81% of the folding mechanisms when tested over the training set of the 63 proteins. It also predicts as a further option the logarithm of the folding rate. To the best of our knowledge, the tool discriminates for the first time whether a protein is characterized by a two state or a multiple state kinetics, during the folding process, and concomitantly estimates also the value of the constant rate of the process. When used to predict the logarithm of the folding rate, K-Fold scores with a correlation value to the experimental data of 0.74 (with a SE of 1.2). AVAILABILITY: http://gpcr.biocomp.unibo.it/cgi/predictors/K-Fold/K-Fold.cgi. SUPPLEMENTARY INFORMATION: http://gpcr.biocomp.unibo.it/~emidio/K-Fold/K-Fold_help.html.

MOTIVATION: Human single nucleotide polymorphisms (SNPs) are the most frequent type of genetic variation in human population. One of the most important goals of SNP projects is to understand which human genotype variations are related to Mendelian and complex diseases. Great interest is focused on non-synonymous coding SNPs (nsSNPs) that are responsible of protein single point mutation. nsSNPs can be neutral or disease associated. It is known that the mutation of only one residue in a protein sequence can be related to a number of pathological conditions of dramatic social impact such as Alzheimer's, Parkinson's and Creutzfeldt-Jakob's diseases. The quality and completeness of presently available SNPs databases allows the application of machine learning techniques to predict the insurgence of human diseases due to single point protein mutation starting from the protein sequence. RESULTS: In this paper, we develop a method based on support vector machines (SVMs) that starting from the protein sequence information can predict whether a new phenotype derived from a nsSNP can be related to a genetic disease in humans. Using a dataset of 21 185 single point mutations, 61% of which are disease-related, out of 3587 proteins, we show that our predictor can reach more than 74% accuracy in the specific task of predicting whether a single point mutation can be disease related or not. Our method, although based on less information, outperforms other web-available predictors implementing different approaches. AVAILABILITY: A beta version of the web tool is available at http://gpcr.biocomp.unibo.it/cgi/predictors/PhD-SNP/PhD-SNP.cgi