212 Hits in 1.3 sec

ExplaiNN: interpretable and transparent neural networks for genomics [article]

German Novakovsky, Oriol Fornes, Manu Saraswat, Sara Mostafavi, Wyeth W Wasserman
2022 bioRxiv   pre-print
Sequence-based deep learning models, particularly convolutional neural networks (CNNs), have shown superior performance on a wide range of genomic tasks. A key limitation of these models is the lack of interpretability, slowing their broad adoption by the genomics community. Current approaches to model interpretation do not readily reveal how a model makes predictions, can be computationally intensive, and depend on the implemented architecture. Here, we introduce ExplaiNN, an adaptation of
more » ... al additive model for genomic tasks wherein predictions are computed as a linear combination of multiple independent CNNs, each consisting of a single convolutional filter and fully connected layers. This approach brings together the expressivity of CNNs with the interpretability of linear models, providing global (cell state level) as well as local (individual sequence level) insights of the biological processes studied. We use ExplaiNN to predict transcription factor (TF) binding and chromatin accessibility states, demonstrating performance levels comparable to state-of-the-art methods, while providing a transparent view of the model's predictions in a straightforward manner. Applied to de novo motif discovery, ExplaiNN detects equivalent motifs to those obtained from specialized algorithms across a range of datasets. Finally, we present ExplaiNN as a plug and play platform in which pre-trained TF binding models and annotated position weight matrices from reference databases can be combined in a simple framework. We expect that ExplaiNN will accelerate the adoption of deep learning by biological domain experts in their daily genomic sequence analyses.
doi:10.1101/2022.05.20.492818 fatcat:2xuif35wene7fbswsi7rdwhwey

Biologically-relevant transfer learning improves transcription factor binding prediction [article]

Gherman Novakovsky, Manu Saraswat, Oriol Fornes, Sara Mostafavi, Wyeth W Wasserman
2020 bioRxiv   pre-print
Background:Deep learning has proven to be a powerful technique for transcription factor (TF) binding prediction, but requires large training datasets. Transfer learning can reduce the amount of data required for deep learning, while improving overall model performance, compared to training a separate model for each new task.Results:We assess a transfer learning strategy for TF binding prediction consisting of a pre-training step, wherein we train a multi-task model with multiple TFs, and a
more » ... tuning step, wherein we initialize single-task models for individual TFs with the weights learned by the multi-task model, after which the single-task models are trained at a lower learning rate. We corroborate that transfer learning improves model performance, especially if in the pre-training step the multi-task model is trained with biologically-relevant TFs. We show the effectiveness of transfer learning for TFs with ~500 ChIP-seq peak regions. Using model interpretation techniques, we demonstrate that the features learned in the pre-training step are refined in the fine-tuning step to resemble the binding motif of the target TF (i.e.the recipient of transfer learning in the fine-tuning step). Moreover, pre-training with biologically-relevant TFs allows single-task models in the fine-tuning step to learn features other than the motif of the target TF.Conclusions:Our results confirm that transfer learning is a powerful technique for TF binding prediction.
doi:10.1101/2020.12.21.423873 fatcat:ijcr22qe4fdordikszv5ypya7q

Cardiovascular Risk Factors in Atrial Fibrillation Associated with Ischaemic Heart Disease

Oriol Yuguero Torres, Jesús Pérez-Mur, Eric Gutiérrez, Joan Valls, Sònia Fornés, Oriol Yuguero Torres
2020 Journal of Integrative Cardiology Open Access  
Objective: To describe cardiovascular risk factors in Atrial Fibrillation (AF) in relation with ischaemic diseases in an emergency service. Methodology: Cross-sectional study of patients with AF attended in the (ES) of the HUAV during 2016. Epidemiological and clinical data and their CVRF were analysed. The statistical association was made through the Chi-Square or Mann-Whitney test. The risk factors associated with AF were adjusted with logistic regression models, calculating OR. Results: We
more » ... aluated 552 patients with 46% men and (54%) women with an average age of 72.9 years. In 57 patients (10.3%), the detection of AF was coincidental. The younger patients presented with more frequent palpitations (p <0.05) and the older patients had dyspnea (p <0.05). The older patients are the ones that take longer to consult (p <0.05). 17% (94) of patients with AF have a heart attack before, during or after the episode of AF, with a higher prevalence among men (p <0.05). The probability of diagnosing ischaemic heart disease in a male patient with AF, hypertensive and diabetic is 71%. Conclusion: In men with hypertension and DM a correct diagnostic and therapeutic management, should consider the diagnostic possibility that AF is related to the presence of ischaemic disease. AF can be considered as an anginal equivalent in patients who meet the three conditions: being male, with hypertension and DM.
doi:10.31487/j.jicoa.2020.06.08 fatcat:jt4r5o2bbvb5jpca46ji2vmjei

ModCRE: a structure homology-modeling approach to predict TF binding in cis-regulatory elements [article]

Oriol Fornes, Alberto Meseguer, Joaquim Aguirre-Plans, PATRICK GOHL, Patricia Mireia-Bota, RUBEN MOLINA-FERNANDEZ, JAUME BONET, ALTAIR CHINCHILLA, FERRAN PEGENAUTE, Oriol Gallego, Narcis Fernandez-Fuentes, Baldo Oliva
2022 bioRxiv   pre-print
Knowledge-based potentials We used the definition of statistical potentials described by Feliu et al. 68 and Fornes et al. 28 .  ... 
doi:10.1101/2022.04.17.488557 fatcat:6qdwaqha3ffltbcmwissnzluiy

Cross-species examination of X-chromosome inactivation highlights domains of escape from silencing [article]

Bradley Philip Balaton, Oriol Fornes, Wyeth W Wasserman, Carolyn J Brown
2020 bioRxiv   pre-print
X-chromosome inactivation (XCI) in eutherian mammals is the epigenetic inactivation of one of the two X chromosomes in XX females in order to compensate for dosage differences with XY males. Not all genes are inactivated, and the proportion escaping from inactivation varies between human and mouse (the two species that have been extensively studied). Results: We used DNA methylation to predict the XCI status of X-linked genes with CpG islands across 12 different species: human, chimp, bonobo,
more » ... rilla, orangutan, mouse, cow, sheep, goat, pig, horse and dog. We determined the XCI status of 342 CpG islands on average per species, with most species having 80-90% of genes subject to XCI. Mouse was an outlier, with a higher proportion of genes subject to XCI than found in other species. Sixteen genes were found to have discordant X-chromosome inactivation statuses across multiple species, with five of these showing primate-specific escape from XCI. These discordant genes tended to cluster together within the X chromosome, along with genes with similar patterns of escape from XCI. CTCF-binding, ATAC-seq signal and LTR repeats were enriched at genes escaping XCI when compared to genes subject to XCI; however, enrichment was only observed in three or four of the species tested. LINE and DNA repeats showed enrichment around subject genes, but again not in a consistent subset of species. Conclusions: In this study we determined XCI status across 12 species, showing mouse to be an outlier with few genes that escape inactivation. Inactivation status is largely conserved across species. The clustering of genes that change XCI status across species implicates a domain-level control. In contrast, the relatively consistent, but not universal correlation of inactivation status with enrichment of repetitive elements or CTCF binding at promoters demonstrates gene-based influences on inactivation state. This study broadens enrichment analysis of regulatory elements to species beyond human and mouse.
doi:10.1101/2020.12.04.412197 fatcat:qscif36vabhhxaliwn5vdnvrwu

Biologically relevant transfer learning improves transcription factor binding prediction

Gherman Novakovsky, Manu Saraswat, Oriol Fornes, Sara Mostafavi, Wyeth W. Wasserman
2021 Genome Biology  
Authors' information Twitter handles: @NovakovskyG (Gherman Novakovsky); @manusaraswat10 (Manu Saraswat); @ofornes (Oriol Fornes); @sara_mostafavi (Sara Mostafavi); @WyWyWa (Wyeth W. Wasserman).  ... 
doi:10.1186/s13059-021-02499-5 pmid:34579793 pmcid:PMC8474956 fatcat:h55prm6hojddfpuxkbjcol46yi

GeneBreaker: Variant simulation to improve the diagnosis of Mendelian rare genetic diseases [article]

Phillip A Richmond, Tamar V Av-Shalom, Oriol Fornes, Bhavi Modi, Wyeth W Wasserman
2020 bioRxiv   pre-print
Mendelian rare genetic diseases affect 5-10% of the population, and with over 5,300 genes responsible for ~7,000 different diseases, they are challenging to diagnose. The use of whole genome sequencing (WGS) for patients and families affected by rare diseases has bolstered the diagnosis rate significantly. Effective use of WGS in this setting relies on the ability to identify the ″broken″ gene responsible for the disease phenotype. This process involves genomic variant calling and
more » ... , and is the beneficiary of rapid improvements to sequencing technology, variant calling approaches, and increased capacity to predict and prioritize genetic variants with potential pathogenicity. As analysis pipelines continue to improve and genomic medicine moves toward a standard of care, careful testing of their efficacy is paramount. However, real-life cases typically emerge anecdotally, and utilization of patient data for the purpose of testing pipeline improvements is regulated and limiting. We identified the need for a gene-based variant simulation framework which can create mock rare disease scenarios, utilizing known pathogenic variants or through the creation of novel gene-disrupting variants. To fill this need, we present GeneBreaker, a tool which creates synthetic rare disease cases with utility for benchmarking variant calling approaches, testing the efficacy of variant prioritization, and as an educational mechanism for the training of diagnostic practitioners in the rapidly growing field of genome medicine. GeneBreaker is freely available at
doi:10.1101/2020.05.29.124495 fatcat:mpimg4p7xfcybhgpuvnxaxxas4

Empathy and burnout of emergency professionals of a health region

Oriol Yuguero, Carles Forné, Montserrat Esquerda, Josep Pifarré, María José Abadías, Joan Viñas
2017 Medicine  
The objective of this study is to assess the association between levels of empathy and burnout of emergency professionals in all the assistance levels. A cross-sectional observational study was conducted in the health region of Lleida and the Pyrenees with 100 professionals from the field of Urgency. Participation reached 40.8%. Empathy and burnout were measured using the Spanish versions of the Jefferson Scale of Physician Empathy (JSPE) and Maslach Burnout Inventory (MBI) respectively. The
more » ... al MBI score and its 3 dimensions (emotional exhaustion, depersonalization, and personal accomplishment) were analyzed. The JSPE and MBI scores were categorized into tertiles that were identified as "low," "moderate," and "high" levels. The median (interquartile range) was 112 (102-123) and 37 (27-53.5) for the JSPE and MBI scores respectively. Professionals with high burnout (MBI≥47) showed the lowest levels of empathy, that is, JSPE score of 105 (98-114); those with moderate burnout (31 MBI < 47) had a JSPE score of 114 (104.5-120.5); and those with low burnout (MBI < 31) had a JSPE score of 120.5 (105.8-127.2). In addition, the highest levels of empathy were associated with the lowest levels of burnout, especially in depersonalization, and to a lesser extent in personal accomplishment. There were no differences in empathy and burnout for any of the other study variables. Our findings suggest that the empathy of emergency professionals is associated with burnout. Hence, reducing professional burnout could help keep emergency professionals' empathy levels high, which in turn would ensure a better quality of care. Nevertheless, it would be necessary to carry out prospective studies to describe the profiles of burnout and empathy as well as their association and evolution. Abbreviations: ED = emergency department, IQR = interquartile range, JSPE = Jefferson Scale of Physician Empathy, MBI = Maslach Burnout Inventory, MBI-DP = depersonalization (dimension of MBI), MBI-EE = emotional exhaustion (dimension of MBI), MBI-PA = personal accomplishment (dimension of MBI), UHAV = University Hospital Arnau de Vilanova.
doi:10.1097/md.0000000000008030 pmid:28906390 pmcid:PMC5604659 fatcat:evknyrtakvc3bbevc6o574p6ve

Multi-class Binary Symbol Classification with Circular Blurred Shape Models [chapter]

Sergio Escalera, Alicia Fornés, Oriol Pujol, Petia Radeva
2009 Lecture Notes in Computer Science  
Multi-class binary symbol classification requires the use of rich descriptors and robust classifiers. Shape representation is a difficult task because of several symbol distortions, such as occlusions, elastic deformations, gaps or noise. In this paper, we present the Circular Blurred Shape Model descriptor. This descriptor encodes the arrangement information of object parts in a correlogram structure. A prior blurring degree defines the level of distortion allowed to the symbol. Moreover, we
more » ... arn the new feature space using a set of Adaboost classifiers, which are combined in the Error-Correcting Output Codes framework to deal with the multi-class categorization problem. The presented work has been validated over different multi-class data sets, and compared to the state-ofthe-art descriptors, showing significant performance improvements.
doi:10.1007/978-3-642-04146-4_107 fatcat:jzgtausckrd23pxicn3luq6j3i

Evaluating the impact of single nucleotide variants on transcription factor binding

Wenqiang Shi, Oriol Fornes, Anthony Mathelier, Wyeth W. Wasserman
2016 Nucleic Acids Research  
Diseases and phenotypes caused by disrupted transcription factor (TF) binding are being identified, but progress is hampered by our limited capacity to predict such functional alterations. Improving predictions may be dependent on expanding the set of bona fide TF binding alterations. Allele-specific binding (ASB) events, where TFs preferentially bind to one of the two alleles at heterozygous sites, reveal the impact of sequence variations in altered TF binding. Here, we present the largest ASB
more » ... compilation to our knowledge, 10 765 ASB events retrieved from 45 ENCODE ChIP-Seq data sets. Our analysis showed that ASB events were frequently associated with motif alterations of the ChIP'ed TF and potential partner TFs, allelic difference of DNase I hypersensitivity and allelic difference of histone modifications. For TF dimers bound symmetrically to DNA, ASB data revealed that central positions of the TF binding motifs were disproportionately important for binding. Lastly, the impact of variation on TF binding was predicted by a classification model incorporating all the investigated features of ASB events. Classification models using only DNase I hypersensitivity and sequence data exhibited predictive accuracy approaching the models with substantially more features. Taken together, the combination of ASB data and the classification model represents an important step toward elucidating regulatory variants across the human genome.
doi:10.1093/nar/gkw691 pmid:27492288 pmcid:PMC5137422 fatcat:k7r43d5iczc3jepu5pedp56h54

Altered transcription factor binding events predict personalized gene expression and confer insight into functional cis-regulatory variants [article]

Wenqiang Shi, Oriol Fornes, Wyeth W Wasserman
2017 bioRxiv   pre-print
Deciphering the functional roles of cis-regulatory variants is a critical challenge in genome analysis and interpretation. We hypothesize that altered transcription factor (TF) binding events are a central mechanism by which cis-regulatory variants impact gene expression. We present TF2Exp, the first gene-based framework (to our knowledge) to predict the impact of altered TF binding on personalized gene expression based on cis-regulatory variants. Using data from lymphoblastoid cell lines,
more » ... p models achieved suitable performance for 3,060 genes. Alterations within DNase I hypersensitive, CTCF-bound, and tissue-specific TF-bound regions were the greatest contributors to the models. Our cis-regulatory variant-based TF2Exp models performed as well as the state-of-the-art SNP-based models, both in cross-validation and external validation. In addition, unlike SNP-based models, our TF2Exp models have the unique advantages to evaluate impact of uncommon variants and distinguish the functional roles of variants in linkage disequilibrium, showing broader utility for future human genetic studies.
doi:10.1101/228155 fatcat:donvrza57bdzll4nyinjcioaxy

Quality of care indicators for a resuscitation unit

Oriol Yuguero, Ana Vena, Carles Forné, Jose Daniel Lacasta, Cecilia Llobet, Maria José Abadías
2018 Medicine  
There are lack of indicators of quality of care in resuscitation units of emergency departments. With the aim of proposing a series of indicators to evaluate the quality of care delivered in hospital resuscitation areas, we conducted a descriptive study of 7579 admissions to the resuscitation unit of an emergency department at a Spanish hospital between 2012 and 2016. The proposed indicators were the percentage of patients attending to the emergency department admitted to the resuscitation area
more » ... by level of triage, the length of stay, the percentage of patients moved to intensive care and surgery at disposition, the mortality in the area or in the emergency department within 24 hours of disposition, and the data completeness. A majority of the patients (62.6%) were men and the median age was 68 years. Over 99% of the required data were recorded. Median length of stay in the resuscitation unit was 0.87 hours (interquartile range, 0.5-1.5). Approximately 80% of patients categorized as an emergency on admission to the emergency department were admitted to the resuscitation unit, although the proportion of urgency patients was higher. The main disposition destination was a trauma cubicle (82.3% of cases). Mortality was 0.41%.Specific indicators are needed to assess the quality of care delivery in resuscitation units. We believe that our findings will provide new insights into the work done to date in this field.
doi:10.1097/md.0000000000013467 pmid:30508973 pmcid:PMC6283204 fatcat:yprcuj7t75avnju2kx727ggjja

RADI (Reduced Alphabet Direct Information): Improving execution time for direct-coupling analysis [article]

Bernat Anton, Mireia Besalu, Oriol Fornes, Jaume Bonet, Gemma De las Cuevas, Narcis Fernandez-Fuentes, Baldomero Oliva
2018 bioRxiv   pre-print
Motivation: Direct-coupling analysis (DCA) for studying the coevolution of residues in proteins has been widely used to predict the three-dimensional structure of a protein from its sequence. Current algorithms for DCA, although efficient, have a high computational cost of determining Direct Information (DI) values for large proteins or domains. In this paper, we present RADI (Reduced Alphabet Direct Information), a variation of the original DCA algorithm that simplifies the computation of DI
more » ... lues by grouping physicochemically equivalent residues. Results: We have compared the first top ranking 40 pairs of DI values and their closest paired contact in 3D. The ranking is also compared with results obtained using a similar but faster approach based on Mutual Information (MI). When we simplify the number of symbols used to describe a protein sequence to 9, RADI achieves similar results as the original DCA (i.e. with the classical alphabet of 21 symbols), while reducing the computation time around 30-fold on large proteins (with length around 1000 residues) and with higher accuracy than predictions based on MI. Interestingly, the simplification produced by grouping amino acids into only two groups (polar and non-polar) is still representative of the physicochemical nature that characterizes the protein structure, having a relevant and useful predictive value, while the computation time is reduced between 100 and 2500-fold. Availability: RADI is available at Supplementary information: Supplementary data is available in the git repository.
doi:10.1101/406603 fatcat:4tbcfrxuivgspaybr5jstxm2eq

A GO catalogue of human DNA-binding transcription factors [article]

Ruth Caroline Lovering, Pascale Gaudet, Marcio Luis Acencio, Alex Ignatchenko, Arttu Jolma, Oriol Fornes, Martin Kuiper, Ivan V Kulakovskiy, Astrid Laegreid, Maria J Martin, Colin Logie
2020 bioRxiv   pre-print
DNA-binding transcription factors recognise genomic addresses, specific sequence motifs in gene regulatory regions, to control gene transcription. A complete and reliable catalogue of all DNA-binding transcription factors is key to investigating the delicate balance of gene regulation in response to environmental and developmental stimuli. The need for such a catalogue of proteins is demonstrated by the many lists of DNA-binding transcription factors that have been produced over the past
more » ... The COST Action Gene Regulation Ensemble Effort for the Knowledge Commons (GREEKC) Consortium brought together experts in the field of transcription with the aim of providing high quality and interoperable gene regulatory data. The Gene Ontology (GO) Consortium provides strict definitions for gene product function, including factors that regulate transcription. The collaboration between the GREEKC and GO Consortia has enabled the application of those definitions to produce a new curated catalogue of human DNA-binding transcription factors, that can be accessed at In addition, this curation effort has led to the GO annotation of almost sixty thousand DNA-binding transcription factors in over a hundred species. Thus, this work will aid researchers investigating the regulation of transcription in both biomedical and basic science.
doi:10.1101/2020.10.28.359232 fatcat:j3joqptjijg5baviwglp2jqmmq

Blurred Shape Model for binary and grey-level symbol recognition

Sergio Escalera, Alicia Fornés, Oriol Pujol, Petia Radeva, Gemma Sánchez, Josep Lladós
2009 Pattern Recognition Letters  
The images have been obtained from original image documents using a semi-supervised segmentation approach (Fornés et al., 2006) .  ...  Data To test the multi-class symbol recognition system, we used three different scenarios: a 7-class handwritten symbols data set, 2 namely clefs and accidentals from old musical scores (Fornés et al.  ... 
doi:10.1016/j.patrec.2009.08.001 fatcat:7pgf25cbcrfwxci5ifkwjtd674
« Previous Showing results 1 — 15 out of 212 results