Optical diagnosis in still images of colorectal polyps: comparison between expert endoscopists and PolyDeep, a Computer-Aided Diagnosis system.

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.

PanDrugs2: prioritizing cancer therapies using integrated individual multi-omics data.

Genomics studies routinely confront researchers with long lists of tumor alterations detected in patients. Such lists are difficult to interpret since only a minority of the alterations are relevant biomarkers for diagnosis and for designing therapeutic strategies. PanDrugs is a methodology that facilitates the interpretation of tumor molecular alterations and guides the selection of personalized treatments. To do so, PanDrugs scores gene actionability and drug feasibility to provide a prioritized evidence-based list of drugs. Here, we introduce PanDrugs2, a major upgrade of PanDrugs that, in addition to somatic variant analysis, supports a new integrated multi-omics analysis which simultaneously combines somatic and germline variants, copy number variation and gene expression data. Moreover, PanDrugs2 now considers cancer genetic dependencies to extend tumor vulnerabilities providing therapeutic options for untargetable genes. Importantly, a novel intuitive report to support clinical decision-making is generated. PanDrugs database has been updated, integrating 23 primary sources that support >74K drug-gene associations obtained from 4642 genes and 14 659 unique compounds. The database has also been reimplemented to allow semi-automatic updates to facilitate maintenance and release of future versions. PanDrugs2 does not require login and is freely available at https://www.pandrugs.org/.

Negative Samples for Improving Object Detection-A Case Study in AI-Assisted Colonoscopy for Polyp Detection.

Deep learning object-detection models are being successfully applied to develop computer-aided diagnosis systems for aiding polyp detection during colonoscopies. Here, we evidence the need to include negative samples for both (i) reducing false positives during the polyp-finding phase, by including images with artifacts that may confuse the detection models (e.g., medical instruments, water jets, feces, blood, excessive proximity of the camera to the colon wall, blurred images, etc.) that are usually not included in model development datasets, and (ii) correctly estimating a more realistic performance of the models. By retraining our previously developed YOLOv3-based detection model with a dataset that includes 15% of additional not-polyp images with a variety of artifacts, we were able to generally improve its F1 performance in our internal test datasets (from an average F1 of 0.869 to 0.893), which now include such type of images, as well as in four public datasets that include not-polyp images (from an average F1 of 0.695 to 0.722).

Performance of Convolutional Neural Networks for Polyp Localization on Public Colonoscopy Image Datasets.

Colorectal cancer is one of the most frequent malignancies. Colonoscopy is the de facto standard for precancerous lesion detection in the colon, i.e., polyps, during screening studies or after facultative recommendation. In recent years, artificial intelligence, and especially deep learning techniques such as convolutional neural networks, have been applied to polyp detection and localization in order to develop real-time CADe systems. However, the performance of machine learning models is very sensitive to changes in the nature of the testing instances, especially when trying to reproduce results for totally different datasets to those used for model development, i.e., inter-dataset testing. Here, we report the results of testing of our previously published polyp detection model using ten public colonoscopy image datasets and analyze them in the context of the results of other 20 state-of-the-art publications using the same datasets. The F1-score of our recently published model was 0.88 when evaluated on a private test partition, i.e., intra-dataset testing, but it decayed, on average, by 13.65% when tested on ten public datasets. In the published research, the average intra-dataset F1-score is 0.91, and we observed that it also decays in the inter-dataset setting to an average F1-score of 0.83.

DREIMT: a drug repositioning database and prioritization tool for immunomodulation.

MOTIVATION: Drug immunomodulation modifies the response of the immune system and can be therapeutically exploited in pathologies such as cancer and autoimmune diseases. RESULTS: DREIMT is a new hypothesis-generation web tool, which performs drug prioritization analysis for immunomodulation. DREIMT provides significant immunomodulatory drugs targeting up to 70 immune cells subtypes through a curated database that integrates 4960 drug profiles and ∼2600 immune gene expression signatures. The tool also suggests potential immunomodulatory drugs targeting user-supplied gene expression signatures. Final output includes drug-signature association scores, FDRs and downloadable plots and results tables. AVAILABILITYAND IMPLEMENTATION: http://www.dreimt.org. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

ATXN1 N-terminal region explains the binding differences of wild-type and expanded forms.

BACKGROUND: Wild-type (wt) polyglutamine (polyQ) regions are implicated in stabilization of protein-protein interactions (PPI). Pathological polyQ expansion, such as that in human Ataxin-1 (ATXN1), that causes spinocerebellar ataxia type 1 (SCA1), results in abnormal PPI. For ATXN1 a larger number of interactors has been reported for the expanded (82Q) than the wt (29Q) protein. METHODS: To understand how the expanded polyQ affects PPI, protein structures were predicted for wt and expanded ATXN1, as well as, for 71 ATXN1 interactors. Then, the binding surfaces of wt and expanded ATXN1 with the reported interactors were inferred. RESULTS: Our data supports that the polyQ expansion alters the ATXN1 conformation and that it enhances the strength of interaction with ATXN1 partners. For both ATXN1 variants, the number of residues at the predicted binding interface are greater after the polyQ, mainly due to the AXH domain. Moreover, the difference in the interaction strength of the ATXN1 variants was due to an increase in the number of interactions at the N-terminal region, before the polyQ, for the expanded form. CONCLUSIONS: There are three regions at the AXH domain that are essential for ATXN1 PPI. The N-terminal region is responsible for the strength of the PPI with the ATXN1 variants. How the predicted motifs in this region affect PPI is discussed, in the context of ATXN1 post-transcriptional modifications.

Predicting Specificities Under the Non-self Gametophytic Self-Incompatibility Recognition Model.

Non-self gametophytic self-incompatibility (GSI) recognition system is characterized by the presence of multiple F-box genes tandemly located in the S-locus, that regulate pollen specificity. This reproductive barrier is present in Solanaceae, Plantaginacea and Maleae (Rosaceae), but only in Petunia functional assays have been performed to get insight on how this recognition mechanism works. In this system, each of the encoded S-pollen proteins (called SLFs in Solanaceae and Plantaginaceae /SFBBs in Maleae) recognizes and interacts with a sub-set of non-self S-pistil proteins, called S-RNases, mediating their ubiquitination and degradation. In Petunia there are 17 SLF genes per S-haplotype, making impossible to determine experimentally each SLF specificity. Moreover, domain -swapping experiments are unlikely to be performed in large scale to determine S-pollen and S-pistil specificities. Phylogenetic analyses of the Petunia SLFs and those from two Solanum genomes, suggest that diversification of SLFs predate the two genera separation. Here we first identify putative SLF genes from nine Solanum and 10 Nicotiana genomes to determine how many gene lineages are present in the three genera, and the rate of origin of new SLF gene lineages. The use of multiple genomes per genera precludes the effect of incompleteness of the genome at the S-locus. The similar number of gene lineages in the three genera implies a comparable effective population size for these species, and number of specificities. The rate of origin of new specificities is one per 10 million years. Moreover, here we determine the amino acids positions under positive selection, those involved in SLF specificity recognition, using 10 Petunia S-haplotypes with more than 11 SLF genes. These 16 amino acid positions account for the differences of self-incompatible (SI) behavior described in the literature. When SLF and S-RNase proteins are divided according to the SI behavior, and the positively selected amino acids classified according to hydrophobicity, charge, polarity and size, we identified fixed differences between SI groups. According to the in silico 3D structure of the two proteins these amino acid positions interact. Therefore, this methodology can be used to infer SLF/S-RNase specificity recognition.

EvoPPI 1.0: a Web Platform for Within- and Between-Species Multiple Interactome Comparisons and Application to Nine PolyQ Proteins Determining Neurodegenerative Diseases.

Protein-protein interaction (PPI) data is essential to elucidate the complex molecular relationships in living systems, and thus understand the biological functions at cellular and systems levels. The complete map of PPIs that can occur in a living organism is called the interactome. For animals, PPI data is stored in multiple databases (e.g., BioGRID, CCSB, DroID, FlyBase, HIPPIE, HitPredict, HomoMINT, INstruct, Interactome3D, mentha, MINT, and PINA2) with different formats. This makes PPI comparisons difficult to perform, especially between species, since orthologous proteins may have different names. Moreover, there is only a partial overlap between databases, even when considering a single species. The EvoPPI ( http://evoppi.i3s.up.pt ) web application presented in this paper allows comparison of data from the different databases at the species level, or between species using a BLAST approach. We show its usefulness by performing a comparative study of the interactome of the nine polyglutamine (polyQ) disease proteins, namely androgen receptor (AR), atrophin-1 (ATN1), ataxin 1 (ATXN1), ataxin 2 (ATXN2), ataxin 3 (ATXN3), ataxin 7 (ATXN7), calcium voltage-gated channel subunit alpha1 A (CACNA1A), Huntingtin (HTT), and TATA-binding protein (TBP). Here we show that none of the human interactors of these proteins is common to all nine interactomes. Only 15 proteins are common to at least 4 of these polyQ disease proteins, and 40% of these are involved in ubiquitin protein ligase-binding function. The results obtained in this study suggest that polyQ disease proteins are involved in different functional networks. Comparisons with Mus musculus PPIs are also made for AR and TBP, using EvoPPI BLAST search approach (a unique feature of EvoPPI), with the goal of understanding why there is a significant excess of common interactors for these proteins in humans.

PanDrugs: a novel method to prioritize anticancer drug treatments according to individual genomic data.

BACKGROUND: Large-sequencing cancer genome projects have shown that tumors have thousands of molecular alterations and their frequency is highly heterogeneous. In such scenarios, physicians and oncologists routinely face lists of cancer genomic alterations where only a minority of them are relevant biomarkers to drive clinical decision-making. For this reason, the medical community agrees on the urgent need of methodologies to establish the relevance of tumor alterations, assisting in genomic profile interpretation, and, more importantly, to prioritize those that could be clinically actionable for cancer therapy. RESULTS: We present PanDrugs, a new computational methodology to guide the selection of personalized treatments in cancer patients using the variant lists provided by genome-wide sequencing analyses. PanDrugs offers the largest database of drug-target associations available from well-known targeted therapies to preclinical drugs. Scoring data-driven gene cancer relevance and drug feasibility PanDrugs interprets genomic alterations and provides a prioritized evidence-based list of anticancer therapies. Our tool represents the first drug prescription strategy applying a rational based on pathway context, multi-gene markers impact and information provided by functional experiments. Our approach has been systematically applied to TCGA patients and successfully validated in a cancer case study with a xenograft mouse model demonstrating its utility. CONCLUSIONS: PanDrugs is a feasible method to identify potentially druggable molecular alterations and prioritize drugs to facilitate the interpretation of genomic landscape and clinical decision-making in cancer patients. Our approach expands the search of druggable genomic alterations from the concept of cancer driver genes to the druggable pathway context extending anticancer therapeutic options beyond already known cancer genes. The methodology is public and easily integratable with custom pipelines through its programmatic API or its docker image. The PanDrugs webtool is freely accessible at http://www.pandrugs.org .

S2P: A software tool to quickly carry out reproducible biomedical research projects involving 2D-gel and MALDI-TOF MS protein data.

BACKGROUND AND OBJECTIVE: 2D-gel electrophoresis is widely used in combination with MALDI-TOF mass spectrometry in order to analyze the proteome of biological samples. For instance, it can be used to discover proteins that are differentially expressed between two groups (e.g. two disease conditions, case vs. control, etc.) thus obtaining a set of potential biomarkers. This procedure requires a great deal of data processing in order to prepare data for analysis or to merge and integrate data from different sources. This kind of work is usually done manually (e.g. copying and pasting data into spreadsheet files), which is highly time consuming and distracts the researcher from other important, core tasks. Moreover, engaging in a repetitive process in a non-automated, handling-based manner is prone to error, thus threatening reliability and reproducibility. The objective of this paper is to present S2P, an open source software to overcome these drawbacks. METHODS: S2P is implemented in Java on top of the AIBench framework, and relies on well-established open source libraries to accomplish different tasks. RESULTS: S2P is an AIBench based desktop multiplatform application, specifically aimed to process 2D-gel and MALDI-mass spectrometry protein identification-based data in a computer-aided, reproducible manner. Different case studies are presented in order to show the usefulness of S2P. CONCLUSIONS: S2P is open source and free to all users at http://www.sing-group.org/s2p. Through its user-friendly GUI interface, S2P dramatically reduces the time that researchers need to invest in order to prepare data for analysis.

Sequential depletion coupled to C18 sequential extraction as a rapid tool for human serum multiple profiling.

Sequential chemical depletion of serum coupled to C18 sequential extraction of peptides as a rapid tool for human serum multiple profiling is herein presented. The methodology comprises depletion with DTT and then with ACN; the extract thus obtained is then summited to fast protein digestion using ultrasonic energy. The pool of peptides is subsequently concentrated using C18-based Zip-tips and the peptides are sequentially extracted using different concentrations of ACN. Each extract is mass-spectrometry profiled with MALDI. The different spectra thus obtained are then successfully used for classification purposes. A total of 40 people, comprising 20 healthy and 20 non-healthy donors, were successfully classified using this method, with an excellent q-value<0.05. The proposed method is cheap as it entails few chemicals, DTT and ACN, simple in terms of handling, and fast. In addition, the methodology is of broad application as it can be used for any study applied to serum samples or other complex biological fluids.

A mesofluidic platform integrating on-chip probe ultrasonication for multiple sample pretreatment involving denaturation, reduction, and digestion in protein identification assays by mass spectrometry.

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.

Fast human serum profiling through chemical depletion coupled to gold-nanoparticle-assisted protein separation.

The use of chemical protein depletion in conjunction with gold-based nanoparticles for fast matrix assisted laser desoption ionization time of flight mass spectrometry-based human serum profiling was assessed. The following variables influencing the process were optimized: (i) amount of nanoparticles, (ii) sample pH, (iii) amount of protein and (iv) incubation time. pH was found the most important factor to be controlled, with an optimum range comprised between 5.8 and 6.4. The minimum incubation time to obtain an adequate profiling was 30 min. Using this approach, serum from five patients with lymphoma, five patients with myeloma and from two healthy volunteers were correctly classified using Principal component analysis.