Carlos Cano

A public resource facilitating clinical use of genomes

Rapid advances in DNA sequencing promise to enable new diagnostics and individualized therapies. Achieving personalized medicine, however, will require extensive research on highly reidentifiable, integrated datasets of genomic and health information. To assist with this, participants in the Personal Genome Project choose to forgo privacy via our institutional review board- approved “open consent” process. The contribution of public data and samples facilitates both scientific discovery and standardization of methods. We present our findings after enrollment of more than 1,800 participants, including whole-genome sequencing of 10 pilot participant genomes (the PGP-10). We introduce the Genome-Environment-Trait Evidence (GET-Evidence) system. This tool automatically processes genomes and prioritizes both published and novel variants for interpretation. In the process of reviewing the presumed healthy PGP-10 genomes, we find numerous literature references implying serious disease. Although it is sometimes impossible to rule out a late-onset effect, stringent evidence requirements can address the high rate of incidental findings. To that end we develop a peer production system for recording and organizing variant evaluations according to standard evidence guidelines, creating a public forum for reaching consensus on interpretation of clinically relevant variants. Genome analysis becomes a two-step process: using a prioritized list to record variant evaluations, then automatically sorting reviewed variants using these annotations. Genome data, health and trait information, participant samples, and variant interpretations are all shared in the public domain—we invite others to review our results using our participant samples and contribute to our interpretations. We offer our public resource and methods to further personalized medical research.

DrugNet: Network-based drug-disease prioritization by integrating heterogeneous data

OBJECTIVE Computational drug repositioning can lead to a considerable reduction in cost and time in any drug development process. Recent approaches have addressed the network-based nature of biological information for performing complex prioritization tasks. In this work, we propose a new methodology based on heterogeneous network prioritization that can aid researchers in the drug repositioning process. METHODS We have developed DrugNet, a new methodology for drug-disease and disease-drug prioritization. Our approach is based on a network-based prioritization method called ProphNet which has recently been developed by the authors. ProphNet is able to integrate data from complex networks involving a wide range of types of elements and interactions. In this work, we built a network of interconnected drugs, proteins and diseases and applied DrugNet to different types of tests for drug repositioning. RESULTS We tested the performance of our approach on different validation tests, including cross validation and tests based on real clinical trials. DrugNet achieved a mean AUC value of 0.9552±0.0015 in 5-fold cross validation tests, and a mean AUC value of 0.8364 for tests based on recent clinical trials (phases 0-4) not present in our data. These results suggest that DrugNet could be very useful for discovering new drug uses. We also studied specific cases of particular interest, proving the benefits of heterogeneous data integration in this problem. CONCLUSIONS Our methodology suggests that new drugs can be repositioned by generating ranked lists of drugs based on a given disease query or vice versa. Our study shows that the simultaneous integration of information about diseases, drugs and targets can lead to a significant improvement in drug repositioning tasks. DrugNet is available as a web tool from http://genome2.ugr.es/drugnet/ (accessed 23.09.14). Matlab source code is also available on the website.

Biomedical application of fuzzy association rules for identifying breast cancer biomarkers

Current breast cancer research involves the study of many different prognosis factors: primary tumor size, lymph node status, tumor grade, tumor receptor status, p53, and ki67 levels, among others. High-throughput microarray technologies are allowing to better understand and identify prognostic factors in breast cancer. But the massive amounts of data derived from these technologies require the use of efficient computational techniques to unveil new and relevant biomedical knowledge. Furthermore, integrative tools are needed that effectively combine heterogeneous types of biomedical data, such as prognosis factors and expression data. The objective of this study was to integrate information from the main prognostic factors in breast cancer with whole-genome microarray data to identify potential associations among them. We propose the application of a data mining approach, called fuzzy association rule mining, to automatically unveil these associations. This paper describes the proposed methodology and illustrates how it can be applied to different breast cancer datasets. The obtained results support known associations involving the number of copies of chromosome-17, HER2 amplification, or the expression level of estrogen and progesterone receptors in breast cancer patients. They also confirm the correspondence between the HER2 status predicted by different testing methodologies (immunohistochemistry and fluorescence in situ hybridization). In addition, other interesting rules involving CDC6, SOX11, and EFEMP1 genes are identified, although further detailed studies are needed to statistically confirm these findings. As part of this study, a web platform implementing the fuzzy association rule mining approach has been made freely available at: http://www.genome2.ugr.es/biofar.

Collaborative text-annotation resource for disease-centered relation extraction from biomedical text

Agglomerating results from studies of individual biological components has shown the potential to produce biomedical discovery and the promise of therapeutic development. Such knowledge integration could be tremendously facilitated by automated text mining for relation extraction in the biomedical literature. Relation extraction systems cannot be developed without substantial datasets annotated with ground truth for benchmarking and training. The creation of such datasets is hampered by the absence of a resource for launching a distributed annotation effort, as well as by the lack of a standardized annotation schema. We have developed an annotation schema and an annotation tool which can be widely adopted so that the resulting annotated corpora from a multitude of disease studies could be assembled into a unified benchmark dataset. The contribution of this paper is threefold. First, we provide an overview of available benchmark corpora and derive a simple annotation schema for specific binary relation extraction problems such as protein-protein and gene-disease relation extraction. Second, we present BioNotate: an open source annotation resource for the distributed creation of a large corpus. Third, we present and make available the results of a pilot annotation effort of the autism disease network.

ProphNet: A generic prioritization method through propagation of information

BackgroundPrioritization methods have become an useful tool for mining large amounts of data to suggest promising hypotheses in early research stages. Particularly, network-based prioritization tools use a network representation for the interactions between different biological entities to identify novel indirect relationships. However, current network-based prioritization tools are strongly tailored to specific domains of interest (e.g. gene-disease prioritization) and they do not allow to consider networks with more than two types of entities (e.g. genes and diseases). Therefore, the direct application of these methods to accomplish new prioritization tasks is limited.ResultsThis work presents ProphNet, a generic network-based prioritization tool that allows to integrate an arbitrary number of interrelated biological entities to accomplish any prioritization task. We tested the performance of ProphNet in comparison with leading network-based prioritization methods, namely rcNet and DomainRBF, for gene-disease and domain-disease prioritization, respectively. The results obtained by ProphNet show a significant improvement in terms of sensitivity and specificity for both tasks. We also applied ProphNet to disease-gene prioritization on Alzheimer, Diabetes Mellitus Type 2 and Breast Cancer to validate the results and identify putative candidate genes involved in these diseases.ConclusionsProphNet works on top of any heterogeneous network by integrating information of different types of biological entities to rank entities of a specific type according to their degree of relationship with a query set of entities of another type. Our method works by propagating information across data networks and measuring the correlation between the propagated values for a query and a target sets of entities. ProphNet is available at: http://genome2.ugr.es/prophnet. A Matlab implementation of the algorithm is also available at the website.

Expression profiling of rectal tumors defines response to neoadjuvant treatment related genes.

To date, no effective method exists that predicts the response to preoperative chemoradiation (CRT) in locally advanced rectal cancer (LARC). Nevertheless, identification of patients who have a higher likelihood of responding to preoperative CRT could be crucial in decreasing treatment morbidity and avoiding expensive and time-consuming treatments. The aim of this study was to identify signatures or molecular markers related to response to pre-operative CRT in LARC. We analyzed the gene expression profiles of 26 pre-treatment biopsies of LARC (10 responders and 16 non-responders) without metastasis using Human WG CodeLink microarray platform. Two hundred and fifty seven genes were differentially over-expressed in the responder patient subgroup. Ingenuity Pathway Analysis revealed a significant ratio of differentially expressed genes related to cancer, cellular growth and proliferation pathways, and c-Myc network. We demonstrated that high Gng4, c-Myc, Pola1, and Rrm1 mRNA expression levels was a significant prognostic factor for response to treatment in LARC patients (p<0.05). Using this gene set, we were able to establish a new model for predicting the response to CRT in rectal cancer with a sensitivity of 60% and 100% specificity. Our results reflect the value of gene expression profiling to gain insight about the molecular pathways involved in the response to treatment of LARC patients. These findings could be clinically relevant and support the use of mRNA levels when aiming to identify patients who respond to CRT therapy.

Intelligent system for the analysis of microarray data using principal components and estimation of distribution algorithms

Background: Microarray technology allows to measure the expression of thousands of genes simultaneously, and under tens of specific conditions. Clustering and Biclustering are the main tools to analyze gene expression data obtained from microarray experiments. By grouping together genes with the same behavior across samples, relevant biological knowledge may be extracted. Non-exclusive groupings are required, since a gene may play more than one biological role. Gene Shaving [Hastie, T., et al. (2000). Gene Shaving as a method for identifying distinct sets of genes with similar expression. Genome Biology, 1, 1-21] is a popular clustering algorithm which looks for coherent clusters of genes with high variance across samples, allowing overlapping among the clusters. Method: In this paper, we present an intelligent system for analyzing microarray data. Our system implements three novel non-exclusive approaches for clustering and biclustering whose aim is to find coherent groups of genes with large between-sample variance: EDA-Clustering and EDA-Biclustering, based on Estimation of Distribution Algorithms (EDA), and Gene-&-Sample Shaving, a biclustering algorithm based on Principal Components Analysis. Results: We integrated the three proposed methods into a web-based platform and tested their performance on two real datasets. The obtained results outperform Gene Shaving in terms of quality and size of revealed patterns. Furthermore, our system allows to visualize the results and validate them from a biological point of view by means of the annotations of the Gene Ontology.

Predictive Biomarkers to Chemoradiation in Locally Advanced Rectal Cancer

There has been a high local recurrence rate in rectal cancer. Besides improvements in surgical techniques, both neoadjuvant short-course radiotherapy and long-course chemoradiation improve oncological results. Approximately 40–60% of rectal cancer patients treated with neoadjuvant chemoradiation achieve some degree of pathologic response. However, there is no effective method of predicting which patients will respond to neoadjuvant treatment. Recent studies have evaluated the potential of genetic biomarkers to predict outcome in locally advanced rectal adenocarcinoma treated with neoadjuvant chemoradiation. The articles produced by the PubMed search were reviewed for those specifically addressing a genetic profiles ability to predict response to neoadjuvant treatment in rectal cancer. Although tissue gene microarray profiling has led to promising data in cancer, to date, none of the identified signatures or molecular markers in locally advanced rectal cancer has been successfully validated as a diagnostic or prognostic tool applicable to routine clinical practice.

Systems Biology as a Comparative Approach to Understand Complex Gene Expression in Neurological Diseases

Systems biology interdisciplinary approaches have become an essential analytical tool that may yield novel and powerful insights about the nature of human health and disease. Complex disorders are known to be caused by the combination of genetic, environmental, immunological or neurological factors. Thus, to understand such disorders, it becomes necessary to address the study of this complexity from a novel perspective. Here, we present a review of integrative approaches that help to understand the underlying biological processes involved in the etiopathogenesis of neurological diseases, for example, those related to autism and autism spectrum disorders (ASD) endophenotypes. Furthermore, we highlight the role of systems biology in the discovery of new biomarkers or therapeutic targets in complex disorders, a key step in the development of personalized medicine, and we demonstrate the role of systems approaches in the design of classifiers that can shorten the time for behavioral diagnosis of autism.

FISim: a new similarity measure between transcription factor binding sites based on the fuzzy integral.

BackgroundRegulatory motifs describe sets of related transcription factor binding sites (TFBSs) and can be represented as position frequency matrices (PFMs). De novo identification of TFBSs is a crucial problem in computational biology which includes the issue of comparing putative motifs with one another and with motifs that are already known. The relative importance of each nucleotide within a given position in the PFMs should be considered in order to compute PFM similarities. Furthermore, biological data are inherently noisy and imprecise. Fuzzy set theory is particularly suitable for modeling imprecise data, whereas fuzzy integrals are highly appropriate for representing the interaction among different information sources.ResultsWe propose FISim, a new similarity measure between PFMs, based on the fuzzy integral of the distance of the nucleotides with respect to the information content of the positions. Unlike existing methods, FISim is designed to consider the higher contribution of better conserved positions to the binding affinity. FISim provides excellent results when dealing with sets of randomly generated motifs, and outperforms the remaining methods when handling real datasets of related motifs. Furthermore, we propose a new cluster methodology based on kernel theory together with FISim to obtain groups of related motifs potentially bound by the same TFs, providing more robust results than existing approaches.ConclusionFISim corrects a design flaw of the most popular methods, whose measures favour similarity of low information content positions. We use our measure to successfully identify motifs that describe binding sites for the same TF and to solve real-life problems. In this study the reliability of fuzzy technology for motif comparison tasks is proven.