RESUMO
BACKGROUND: In return for their nutritional properties and broad availability, cereal crops have been associated with different alimentary disorders and symptoms, with the majority of the responsibility being attributed to gluten. Therefore, the research of gluten-related literature data continues to be produced at ever-growing rates, driven in part by the recent exploratory studies that link gluten to non-traditional diseases and the popularity of gluten-free diets, making it increasingly difficult to access and analyse practical and structured information. In this sense, the accelerated discovery of novel advances in diagnosis and treatment, as well as exploratory studies, produce a favourable scenario for disinformation and misinformation. OBJECTIVES: Aligned with, the European Union strategy "Delivering on EU Food Safety and Nutrition in 2050â³ which emphasizes the inextricable links between imbalanced diets, the increased exposure to unreliable sources of information and misleading information, and the increased dependency on reliable sources of information; this paper presents GlutKNOIS, a public and interactive literature-based database that reconstructs and represents the experimental biomedical knowledge extracted from the gluten-related literature. The developed platform includes different external database knowledge, bibliometrics statistics and social media discussion to propose a novel and enhanced way to search, visualise and analyse potential biomedical and health-related interactions in relation to the gluten domain. METHODS: For this purpose, the presented study applies a semi-supervised curation workflow that combines natural language processing techniques, machine learning algorithms, ontology-based normalization and integration approaches, named entity recognition methods, and graph knowledge reconstruction methodologies to process, classify, represent and analyse the experimental findings contained in the literature, which is also complemented by data from the social discussion. RESULTS AND CONCLUSIONS: In this sense, 5814 documents were manually annotated and 7424 were fully automatically processed to reconstruct the first online gluten-related knowledge database of evidenced health-related interactions that produce health or metabolic changes based on the literature. In addition, the automatic processing of the literature combined with the knowledge representation methodologies proposed has the potential to assist in the revision and analysis of years of gluten research. The reconstructed knowledge base is public and accessible at https://sing-group.org/glutknois/.
Assuntos
Glutens , Bases de Conhecimento , Humanos , Aprendizado de Máquina , Algoritmos , Processamento de Linguagem NaturalRESUMO
Motivated by the pervasiveness of artificial intelligence (AI) and the Internet of Things (IoT) in the current "smart everything" scenario, this article provides a comprehensive overview of the most recent research at the intersection of both domains, focusing on the design and development of specific mechanisms for enabling a collaborative inference across edge devices towards the in situ execution of highly complex state-of-the-art deep neural networks (DNNs), despite the resource-constrained nature of such infrastructures. In particular, the review discusses the most salient approaches conceived along those lines, elaborating on the specificities of the partitioning schemes and the parallelism paradigms explored, providing an organized and schematic discussion of the underlying workflows and associated communication patterns, as well as the architectural aspects of the DNNs that have driven the design of such techniques, while also highlighting both the primary challenges encountered at the design and operational levels and the specific adjustments or enhancements explored in response to them.
RESUMO
Developing models able to predict interactions between drugs and enzymes is a primary goal in computational biology since these models may be used for predicting both new active drugs and the interactions between known drugs on untested targets. With the compilation of a large dataset of drug-enzyme pairs (62,524), we recognized a unique opportunity to attempt to build a novel multi-target machine learning (MTML) quantitative structure-activity relationship (QSAR) model for probing interactions among different drugs and enzyme targets. To this end, this paper presents an MTML-QSAR model based on using the features of topological drugs together with the artificial neural network (ANN) multi-layer perceptron (MLP). Validation of the final best model found was carried out by internal cross-validation statistics and other relevant diagnostic statistical parameters. The overall accuracy of the derived model was found to be higher than 96%. Finally, to maximize the diffusion of this model, a public and accessible tool has been developed to allow users to perform their own predictions. The developed web-based tool is public accessible and can be downloaded as free open-source software.
Assuntos
Relação Quantitativa Estrutura-Atividade , Software , Redes Neurais de Computação , Aprendizado de Máquina , InternetRESUMO
The human gut microbiome impacts several aspects of human health and disease, including digestion, drug metabolism and the propensity to develop various inflammatory, autoimmune and metabolic diseases. Many of the molecular processes that play a role in the activity and dynamics of the microbiota go beyond species and genic composition and thus, their understanding requires advanced bioinformatics support. This article aims to provide an up-to-date view of the resources and software tools that are being developed and used in human gut microbiome research, in particular data integration and systems-level analysis efforts. These efforts demonstrate the power of standardized and reproducible computational workflows for integrating and analysing varied omics data and gaining deeper insights into microbe community structure and function as well as host-microbe interactions.
Assuntos
Microbioma Gastrointestinal , Biologia Computacional , Ensaios de Triagem em Larga Escala , Humanos , Reprodutibilidade dos Testes , SoftwareRESUMO
BACKGROUND: L-ascorbate (Vitamin C) is an important antioxidant and co-factor in eukaryotic cells, and in mammals it is indispensable for brain development and cognitive function. Vertebrates usually become L-ascorbate auxothrophs when the last enzyme of the synthetic pathway, an L-gulonolactone oxidase (GULO), is lost. Since Protostomes were until recently thought not to have a GULO gene, they were considered to be auxothrophs for Vitamin C. RESULTS: By performing phylogenetic analyses with tens of non-Bilateria and Protostomian genomes, it is shown, that a GULO gene is present in the non-Bilateria Placozoa, Myxozoa (here reported for the first time) and Anthozoa groups, and in Protostomians, in the Araneae family, the Gastropoda class, the Acari subclass (here reported for the first time), and the Priapulida, Annelida (here reported for the first time) and Brachiopoda phyla lineages. GULO is an old gene that predates the separation of Animals and Fungi, although it could be much older. We also show that within Protostomes, GULO has been lost multiple times in large taxonomic groups, namely the Pancrustacea, Nematoda, Platyhelminthes and Bivalvia groups, a pattern similar to that reported for Vertebrate species. Nevertheless, we show that Drosophila melanogaster seems to be capable of synthesizing L-ascorbate, likely through an alternative pathway, as recently reported for Caenorhabditis elegans. CONCLUSIONS: Non-Bilaterian and Protostomians seem to be able to synthesize Vitamin C either through the conventional animal pathway or an alternative pathway, but in this animal group, not being able to synthesize L-ascorbate seems to be the exception rather than the rule.
Assuntos
Ácido Ascórbico/metabolismo , Eucariotos/enzimologia , Eucariotos/genética , Evolução Molecular , L-Gulonolactona Oxidase/genética , Animais , Drosophila melanogaster/genética , Eucariotos/classificação , Eucariotos/metabolismo , Genoma , L-Gulonolactona Oxidase/química , L-Gulonolactona Oxidase/metabolismo , Modelos Moleculares , Filogenia , Vertebrados/classificação , Vertebrados/genéticaRESUMO
Summary: High-throughput sequencing of bisulfite-converted DNA is a technique used to measure DNA methylation levels. Although a considerable number of computational pipelines have been developed to analyze such data, none of them tackles all the peculiarities of the analysis together, revealing limitations that can force the user to manually perform additional steps needed for a complete processing of the data. This article presents bicycle, an integrated, flexible analysis pipeline for bisulfite sequencing data. Bicycle analyzes whole genome bisulfite sequencing data, targeted bisulfite sequencing data and hydroxymethylation data. To show how bicycle overtakes other available pipelines, we compared them on a defined number of features that are summarized in a table. We also tested bicycle with both simulated and real datasets, to show its level of performance, and compared it to different state-of-the-art methylation analysis pipelines. Availability and implementation: Bicycle is publicly available under GNU LGPL v3.0 license at http://www.sing-group.org/bicycle. Users can also download a customized Ubuntu LiveCD including bicycle and other bisulfite sequencing data pipelines compared here. In addition, a docker image with bicycle and its dependencies, which allows a straightforward use of bicycle in any platform (e.g. Linux, OS X or Windows), is also available. Contact: ograna@cnio.es or dgpena@uvigo.es. Supplementary information: Supplementary data are available at Bioinformatics online.
Assuntos
Metilação de DNA , Sequenciamento de Nucleotídeos em Larga Escala/métodos , Análise de Sequência de DNA/métodos , Software , Biologia Computacional , SulfitosRESUMO
Advances in the field of genome sequencing have enabled a comprehensive analysis and annotation of the dynamics of the protein inventory of cells. This has been proven particularly rewarding for microbial cells, for which the majority of proteins are already accessible to analysis through automatic metagenome annotation. The large-scale in silico screening of proteomes and metaproteomes is key to uncover bioactivities of translational, clinical and biotechnological interest, and to help assign functions to certain proteins, such as those predicted as hypothetical. This work introduces a new method for the prediction of the bioactivity potential of proteomes/metaproteomes, supporting the discovery of functionally relevant proteins based on prior knowledge. This methodology complements functional annotation enrichment methods by allowing the assignment of functions to proteins annotated as hypothetical/putative/uncharacterised, as well as and enabling the detection of specific bioactivities and the recovery of proteins from defined taxa. This work shows how the new method can be applied to screen proteome and metaproteome sets to obtain predictions of clinical or biotechnological interest based on reference datasets. Notably, with this methodology, the large information files obtained after DNA sequencing or protein identification experiments can be associated for translational purposes that, in cases such as antibiotic-resistance pathogens or foodborne diseases, may represent changes in how these important and global health burdens are approached in the clinical practice. Finally, the Sequence-based Expert-driven pRoteome bioactivity Prediction EnvironmENT, a public Web service implemented in Scala functional programming style, is introduced as means to ensure broad access to the method as well as to discuss main implementation issues, such as modularity, extensibility and interoperability.
Assuntos
Biologia Computacional/métodos , Proteoma , InternetRESUMO
Peptidome similarity analysis enables researchers to gain insights into differential peptide profiles, providing a robust tool to discriminate strain-specific peptides, true intra-species differences among biological replicates or even microorganism-phenotype variations. However, no in silico peptide fingerprinting software existed to facilitate such phylogeny inference. Hence, we developed the Peptidomes for Phylogenies (P4P) web tool, which enables the survey of similarities between microbial proteomes and simplifies the process of obtaining new biological insights into their phylogeny. P4P can be used to analyze different peptide datasets, i.e. bacteria, viruses, eukaryotic species or even metaproteomes. Also, it is able to work with whole proteome datasets and experimental mass-to-charge lists originated from mass spectrometers. The ultimate aim is to generate a valid and manageable list of peptides that have phylogenetic signal and are potentially sample-specific. Sample-to-sample comparison is based on a consensus peak set matrix, which can be further submitted to phylogenetic analysis. P4P holds great potential for improving phylogenetic analyses in challenging taxonomic groups, biomarker identification or epidemiologic studies. Notably, P4P can be of interest for applications handling large proteomic datasets, which it is able to reduce to small matrices while maintaining high phylogenetic resolution. The web server is available at http://sing-group.org/p4p.
Assuntos
Bactérias/classificação , Mapeamento de Peptídeos , Filogenia , Proteômica , Software , Bacillus cereus/classificação , Bacillus cereus/genética , Bactérias/genética , Bifidobacterium animalis/classificação , Bifidobacterium animalis/genética , Internet , Peptídeos/análise , Peptídeos/química , Proteoma , Ralstonia solanacearum/classificação , Ralstonia solanacearum/genéticaRESUMO
BACKGROUND: Nowadays, the use of social media is part of daily life, with more and more people, including governments and health organizations, using at least one platform regularly. Social media enables users to interact among large groups of people that share the same interests and suffer the same afflictions. Notably, these channels promote the ability to find and share information about health and medical conditions. OBJECTIVE: This study aimed to characterize the bowel disease (BD) community on Twitter, in particular how patients understand, discuss, feel, and react to the condition. The main questions were as follows: Which are the main communities and most influential users?; Where are the main content providers from?; What are the key biomedical and scientific topics under discussion? How are topics interrelated in patient communications?; How do external events influence user activity?; What kind of external sources of information are being promoted? METHODS: To answer these questions, a dataset of tweets containing terms related to BD conditions was collected from February to August 2018, accounting for a total of 24,634 tweets from 13,295 different users. Tweet preprocessing entailed the extraction of textual contents, hyperlinks, hashtags, time, location, and user information. Missing and incomplete information about the user profiles was completed using different analysis techniques. Semantic tweet topic analysis was supported by a lexicon-based entity recognizer. Furthermore, sentiment analysis enabled a closer look into the opinions expressed in the tweets, namely, gaining a deeper understanding of patients' feelings and experiences. RESULTS: Health organizations received most of the communication, whereas BD patients and experts in bowel conditions and nutrition were among those tweeting the most. In general, the BD community was mainly discussing symptoms, BD-related diseases, and diet-based treatments. Diarrhea and constipation were the most commonly mentioned symptoms, and cancer, anxiety disorder, depression, and chronic inflammations were frequently part of BD-related tweets. Most patient tweets discussed the bad side of BD conditions and other related conditions, namely, depression, diarrhea, and fibromyalgia. In turn, gluten-free diets and probiotic supplements were often mentioned in patient tweets expressing positive emotions. However, for the most part, tweets containing mentions to foods and diets showed a similar distribution of negative and positive sentiments because the effects of certain food components (eg, fiber, iron, and magnesium) were perceived differently, depending on the state of the disease and other personal conditions of the patients. The benefits of medical cannabis for the treatment of different chronic diseases were also highlighted. CONCLUSIONS: This study evidences that Twitter is becoming an influential space for conversation about bowel conditions, namely, patient opinions about associated symptoms and treatments. So, further qualitative and quantitative content analyses hold the potential to support decision making among health-related stakeholders, including the planning of awareness campaigns.
Assuntos
Neoplasias do Colo/psicologia , Conhecimentos, Atitudes e Prática em Saúde , Síndrome do Intestino Irritável/psicologia , Mídias Sociais , Inquéritos e Questionários , Demografia , Saúde Global , HumanosRESUMO
Experimental incapacity to track microbe-microbe interactions in structures like biofilms, and the complexity inherent to the mathematical modelling of those interactions, raises the need for feasible, alternative modelling approaches. This work proposes an agent-based representation of the diffusion of N-acyl homoserine lactones (AHL) in a multicellular environment formed by Pseudomonas aeruginosa and Candida albicans. Depending on the spatial location, C. albicans cells were variably exposed to AHLs, an observation that might help explain why phenotypic switching of individual cells in biofilms occurred at different time points. The simulation and algebraic results were similar for simpler scenarios, although some statistical differences could be observed (p < 0.05). The model was also successfully applied to a more complex scenario representing a small multicellular environment containing C. albicans and P. aeruginosa cells encased in a 3-D matrix. Further development of this model may help create a predictive tool to depict biofilm heterogeneity at the single-cell level.
Assuntos
Acil-Butirolactonas/química , Candida albicans/metabolismo , Modelos Teóricos , Pseudomonas aeruginosa/metabolismo , Percepção de Quorum , Biofilmes , Candida albicans/fisiologia , Difusão , Pseudomonas aeruginosa/fisiologiaRESUMO
A novel ecosystem to promote the physical, emotional and psychic health and well-being of the elderly is presented. Our proposal was designed to add several services developed to meet the needs of the senior population, namely services to improve social inclusion and increase contribution to society. Moreover, the solution monitors the vital signs of elderly individuals, as well as environmental parameters and behavior patterns, in order to seek eminent danger situations and predict potential hazardous issues, acting in accordance with the various alert levels specified for each individual. The platform was tested by seniors in a real scenario. The experimental results demonstrated that the proposed ecosystem was well accepted and is easy to use by seniors.
Assuntos
Ecossistema , Idoso , Atenção à Saúde , Humanos , TelemedicinaRESUMO
Typical bacterial strain differentiation methods are often challenged by high genetic similarity between strains. To address this problem, we introduce a novel in silico peptide fingerprinting method based on conventional wet-lab protocols that enables the identification of potential strain-specific peptides. These can be further investigated using in vitro approaches, laying a foundation for the development of biomarker detection and application-specific methods. This novel method aims at reducing large amounts of comparative peptide data to binary matrices while maintaining a high phylogenetic resolution. The underlying case study concerns the Bacillus cereus group, namely the differentiation of Bacillus thuringiensis, Bacillus anthracis and Bacillus cereus strains. Results show that trees based on cytoplasmic and extracellular peptidomes are only marginally in conflict with those based on whole proteomes, as inferred by the established Genome-BLAST Distance Phylogeny (GBDP) method. Hence, these results indicate that the two approaches can most likely be used complementarily even in other organismal groups. The obtained results confirm previous reports about the misclassification of many strains within the B. cereus group. Moreover, our method was able to separate the B. anthracis strains with high resolution, similarly to the GBDP results as benchmarked via Bayesian inference and both Maximum Likelihood and Maximum Parsimony. In addition to the presented phylogenomic applications, whole-peptide fingerprinting might also become a valuable complementary technique to digital DNA-DNA hybridization, notably for bacterial classification at the species and subspecies level in the future.
Assuntos
Bacillus/classificação , Proteínas de Bactérias/classificação , Peptídeos/classificação , Proteoma/classificação , Proteômica/métodos , Bacillus/genética , Proteínas de Bactérias/genética , DNA Bacteriano/genética , Bases de Dados de Proteínas , Modelos Genéticos , Peptídeos/genética , Filogenia , Proteoma/genética , Especificidade da EspécieRESUMO
In this work we present the design and implementation of WARCProcessor, a novel multiplatform integrative tool aimed to build scientific datasets to facilitate experimentation in web spam research. The developed application allows the user to specify multiple criteria that change the way in which new corpora are generated whilst reducing the number of repetitive and error prone tasks related with existing corpus maintenance. For this goal, WARCProcessor supports up to six commonly used data sources for web spam research, being able to store output corpus in standard WARC format together with complementary metadata files. Additionally, the application facilitates the automatic and concurrent download of web sites from Internet, giving the possibility of configuring the deep of the links to be followed as well as the behaviour when redirected URLs appear. WARCProcessor supports both an interactive GUI interface and a command line utility for being executed in background.
RESUMO
Web services are the de facto standard in biomedical data integration. However, there are data integration scenarios that cannot be fully covered by Web services. A number of Web databases and tools do not support Web services, and existing Web services do not cover for all possible user data demands. As a consequence, Web data scraping, one of the oldest techniques for extracting Web contents, is still in position to offer a valid and valuable service to a wide range of bioinformatics applications, ranging from simple extraction robots to online meta-servers. This article reviews existing scraping frameworks and tools, identifying their strengths and limitations in terms of extraction capabilities. The main focus is set on showing how straightforward it is today to set up a data scraping pipeline, with minimal programming effort, and answer a number of practical needs. For exemplification purposes, we introduce a biomedical data extraction scenario where the desired data sources, well-known in clinical microbiology and similar domains, do not offer programmatic interfaces yet. Moreover, we describe the operation of WhichGenes and PathJam, two bioinformatics meta-servers that use scraping as means to cope with gene set enrichment analysis.
Assuntos
Internet , Sistemas de Gerenciamento de Base de Dados , Interface Usuário-ComputadorRESUMO
Bifidobacteria are gut commensal microorganisms belonging to the Actinobacteria group. Some specific strains of Bifidobacterium animalis subsp. lactis are used in functional foods as they are able to exert health-promoting effects in the human host. Due to the limited genetic variability within this subspecies, it is sometimes difficult for a manufacturer to properly track its strain once included in dairy products or functional foods. In this paper, we present a peptidome-based analysis in which the proteomes of a set of B. animalis subsp. lactis strains were digested in silico with human gut endopeptidases. The molecular masses were compared along all the strains to detect strain-specific peptides. These peptides may be interesting towards the development of methodologies for strain identification in the final product.
Assuntos
Proteínas de Bactérias/análise , Bifidobacterium animalis/química , Bifidobacterium animalis/isolamento & purificação , Peptídeos/análise , Proteoma/análise , Proteínas de Bactérias/genética , Bifidobacterium animalis/classificação , Bifidobacterium animalis/genética , Simulação por Computador , Laticínios/microbiologia , Endopeptidases/química , Genoma Bacteriano , Humanos , Peptídeos/isolamento & purificação , Filogenia , Proteômica/métodos , Análise de Sequência de DNARESUMO
BACKGROUND: The diagnosis and prognosis of several diseases can be shortened through the use of different large-scale genome experiments. In this context, microarrays can generate expression data for a huge set of genes. However, to obtain solid statistical evidence from the resulting data, it is necessary to train and to validate many classification techniques in order to find the best discriminative method. This is a time-consuming process that normally depends on intricate statistical tools. RESULTS: geneCommittee is a web-based interactive tool for routinely evaluating the discriminative classification power of custom hypothesis in the form of biologically relevant gene sets. While the user can work with different gene set collections and several microarray data files to configure specific classification experiments, the tool is able to run several tests in parallel. Provided with a straightforward and intuitive interface, geneCommittee is able to render valuable information for diagnostic analyses and clinical management decisions based on systematically evaluating custom hypothesis over different data sets using complementary classifiers, a key aspect in clinical research. CONCLUSIONS: geneCommittee allows the enrichment of microarrays raw data with gene functional annotations, producing integrated datasets that simplify the construction of better discriminative hypothesis, and allows the creation of a set of complementary classifiers. The trained committees can then be used for clinical research and diagnosis. Full documentation including common use cases and guided analysis workflows is freely available at http://sing.ei.uvigo.es/GC/.
Assuntos
Biologia Computacional/métodos , Bases de Dados Genéticas , Perfilação da Expressão Gênica/métodos , Internet , Software , Doença/genética , Humanos , Interface Usuário-ComputadorRESUMO
The integration of ultrasound (US)-assisted sample processing on-chip in a lab-on-a-valve (LOV) format for automated high-throughput shotgun proteomic assays is herein presented for the first time. The proof of concept of this system was demonstrated with the analysis of three proteins and sera from patients with lymphoma or myeloma.
Assuntos
Biomarcadores Tumorais/análise , Espectrometria de Massas/métodos , Procedimentos Analíticos em Microchip/métodos , Técnicas Analíticas Microfluídicas/métodos , Desnaturação Proteica , HumanosRESUMO
The low average birth rate in developed countries and the increase in life expectancy have lead society to face for the first time an ageing situation. This situation associated with the World's economic crisis (which started in 2008) forces the need of equating better and more efficient ways of providing more quality of life for the elderly. In this context, the solution presented in this work proposes to tackle the problem of monitoring the elderly in a way that is not restrictive for the life of the monitored, avoiding the need for premature nursing home admissions. To this end, the system uses the fusion of sensory data provided by a network of wireless sensors placed on the periphery of the user. Our approach was also designed with a low-cost deployment in mind, so that the target group may be as wide as possible. Regarding the detection of long-term problems, the tests conducted showed that the precision of the system in identifying and discerning body postures and body movements allows for a valid monitorization and rehabilitation of the user. Moreover, concerning the detection of accidents, while the proposed solution presented a near 100% precision at detecting normal falls, the detection of more complex falls (i.e., hampered falls) will require further study.
Assuntos
Acidentes por Quedas , Monitorização Ambulatorial , Telemetria/métodos , Tecnologia sem Fio , Acelerometria/instrumentação , Acelerometria/métodos , Adulto , Estatura , Peso Corporal , Feminino , Humanos , Masculino , Monitorização Ambulatorial/instrumentação , Movimento , Postura , Qualidade de Vida , Telemetria/economiaRESUMO
Ambient Intelligence promises to transform current spaces into electronic environments that are responsive, assistive and sensitive to human presence. Those electronic environments will be fully populated with dozens, hundreds or even thousands of connected devices that share information and thus become intelligent. That massive wave of electronic devices will also invade everyday objects, turning them into smart entities, keeping their native features and characteristics while seamlessly promoting them to a new class of thinking and reasoning everyday objects. Although there are strong expectations that most of the users' needs can be fulfilled without their intervention, there are still situations where interaction is required. This paper presents work being done in the field of human-computer interaction, focusing on smart home environments, while being a part of a larger project called Aging Inside a Smart Home. This initiative arose as a way to deal with a large scourge of our country, where lots of elderly persons live alone in their homes, often with limited or no physical mobility. The project relies on the mobile agent computing paradigm in order to create a Virtual Butler that provides the interface between the elderly and the smart home infrastructure. The Virtual Butler is receptive to user questions, answering them according to the context and knowledge of the AISH. It is also capable of interacting with the user whenever it senses that something has gone wrong, notifying next of kin and/or medical services, etc. The Virtual Butler is aware of the user location and moves to the computing device which is closest to the user, in order to be always present. Its avatar can also run in handheld devices keeping its main functionality in order to track user when s/he goes out. According to the evaluation carried out, the Virtual Butler is assessed as a very interesting and loved digital friend, filling the gap between the user and the smart home. The evaluation also showed that the Virtual Butler concept can be easily ported to other types of possible smart and assistive environments like airports, hospitals, shopping malls, offices, etc.
Assuntos
Inteligência Artificial , Interface Usuário-Computador , Idoso , Meio Ambiente , Humanos , SoftwareRESUMO
Background: PolyDeep is a computer-aided detection and classification (CADe/x) system trained to detect and classify polyps. During colonoscopy, CADe/x systems help endoscopists to predict the histology of colonic lesions. Objective: To compare the diagnostic performance of PolyDeep and expert endoscopists for the optical diagnosis of colorectal polyps on still images. Methods: PolyDeep Image Classification (PIC) is an in vitro diagnostic test study. The PIC database contains NBI images of 491 colorectal polyps with histological diagnosis. We evaluated the diagnostic performance of PolyDeep and four expert endoscopists for neoplasia (adenoma, sessile serrated lesion, traditional serrated adenoma) and adenoma characterization and compared them with the McNemar test. Receiver operating characteristic curves were constructed to assess the overall discriminatory ability, comparing the area under the curve of endoscopists and PolyDeep with the chi- square homogeneity areas test. Results: The diagnostic performance of the endoscopists and PolyDeep in the characterization of neoplasia is similar in terms of sensitivity (PolyDeep: 89.05%; E1: 91.23%, p=0.5; E2: 96.11%, p<0.001; E3: 86.65%, p=0.3; E4: 91.26% p=0.3) and specificity (PolyDeep: 35.53%; E1: 33.80%, p=0.8; E2: 34.72%, p=1; E3: 39.24%, p=0.8; E4: 46.84%, p=0.2). The overall discriminative ability also showed no statistically significant differences (PolyDeep: 0.623; E1: 0.625, p=0.8; E2: 0.654, p=0.2; E3: 0.629, p=0.9; E4: 0.690, p=0.09). In the optical diagnosis of adenomatous polyps, we found that PolyDeep had a significantly higher sensitivity and a significantly lower specificity. The overall discriminative ability of adenomatous lesions by expert endoscopists is significantly higher than PolyDeep (PolyDeep: 0.582; E1: 0.685, p < 0.001; E2: 0.677, p < 0.0001; E3: 0.658, p < 0.01; E4: 0.694, p < 0.0001). Conclusion: PolyDeep and endoscopists have similar diagnostic performance in the optical diagnosis of neoplastic lesions. However, endoscopists have a better global discriminatory ability than PolyDeep in the optical diagnosis of adenomatous polyps.