Proceedings of the 15th Annual UT-KBRIN Bioinformatics Summit 2016: Cadiz, KY, USA. 8-10 April 2016

Eric C. Rouchka, Julia H. Chariker, Benjamin J. Harrison, Juw Won Park, Xueyuan Cao, Stanley Pounds, Susana C. Raimondi, James R. Downing, Raul C. Ribeiro, Jeffrey Rubnitz, Jatinder Lamba, Bernie J. Daigle Jr., Deborah Burgess, Stephanie Gehrlich, John C. Carmen, Nicholas Johnson, Chandrakanth Emani, Kalpani De Silva, Michael P. Heaton, Theodore S. Kalbfleisch, Teeradache Viangteeravat, Rahul Mudunuri, Oluwaseun Ajayi, Fatih Sen, Eunice Y. Huang, Mohammad Mohebbi, Luaire Florian, Douglas J. Jackson, John F. Naber, Akm Sabbir, Sally R. Ellingson, Yuping Lu, Charles A. Phillips, Michael A. Langston, Rahul Kumar Sevakula, Raghuveer Thirukovalluru, Nishchal K. Verma, Yan Cui, Mohammed Sayed, Jing Wang, Qi Liu, Yu Shyr, Xiaofei Zhang, Naresh Prodduturi, Gavin R. Oliver, Diane Grill, Jie Na, Jeanette Eckel-Passow, Eric W. Klee, Michael M. Goodin, Mark L. Farman, Harrison Inocencio, Chanyong Jang, Jerzy W. Jaromczyk, Neil Moore, Kelly L. Sovacool, Leon Dent, Mike Izban, Sammed N. Mandape, Shruti S. Sakhare, Siddharth Pratap, Dana Marshall, M. Scotty Depriest, James N. MacLeod, Hanady Adam, Ethan Blandford, Joel Campbell, Joshua Castlen, Brittany Dixon, Ginger Gilbert, Aaron Hall, Philip Kreisle, Jessica Lasher, Bethany Oakes, Allison Speer, Maximilian Valentine, Naga Satya Venkateswara Ra Nagisetty, Rony Jose, Robert W. Rooney, David HainsBMC Bioinformatics 2016 · DOI

TL;DR

This document summarizes the proceedings of the 15th Annual UT-KBRIN Bioinformatics Summit, showcasing a diverse range of research in bioinformatics. The summit highlighted advancements in areas like genomic data analysis, disease modeling, and the development of novel computational tools, emphasizing the growing importance of interdisciplinary collaboration in tackling complex biological questions.

What Problem Does This Paper Address?

The collection of papers presented at the 15th Annual UT-KBRIN Bioinformatics Summit addresses a broad spectrum of challenges within the field of bioinformatics. These include:

  • Improving the efficiency and accuracy of genomic data analysis: Many studies focus on developing new algorithms or refining existing ones for tasks like variant calling, gene expression analysis, and motif discovery, aiming to handle the ever-increasing volume and complexity of genomic data.
  • Understanding the biological mechanisms of disease: Researchers are exploring how to integrate diverse data types (genomic, transcriptomic, epigenomic, clinical) to gain deeper insights into disease pathways, identify potential biomarkers, and predict patient outcomes.
  • Developing novel computational tools and platforms: The summit features the creation of new software and systems designed to facilitate research, such as tools for in silico genome comparison, data integration platforms for biorepositories, and methods for tracking molecular evolution.
  • Bridging the gap between computational findings and clinical applications: A recurring theme is the translation of bioinformatics research into practical applications, whether it's for personalized medicine, drug discovery, or improved diagnostic tests.

Methods

The summit showcases a wide array of bioinformatics methodologies, reflecting the field's interdisciplinary nature. Key techniques and approaches include:

  • Genomic Data Analysis:
    • UCSC Genome Browser: Used for visualization and exploration of genomic annotation tracks.
  • Genome Analysis Tool Kit (GATK): Employed for genomic variation detection and pipeline benchmarking.
  • Next-Generation Sequencing (NGS) analysis: Including RNA-Seq data processing and analysis of whole-genome shotgun sequencing data.
  • BLAST: Utilized for sequence alignment and identifying homologous regions.
  • Machine Learning and Statistical Modeling:
    • Bayesian networks: Applied in approaches like NOODLE for integrating miRNA-mRNA interaction networks.
  • Deep Learning: Specifically, stacked denoising sparse auto-encoders (SDSAE) for transcriptome-based cancer classification.
  • Symbolic Aggregate Approximation (SAX) and Piecewise Aggregate Approximation (PAA): Used for converting physiological data into symbolic patterns for data mining.
  • Low Rank Matrix Decomposition (LRMD): Applied to derive patient similarity matrices from physiological data.
  • K-means clustering: Used for motif discovery in DNA sequences.
  • Canonical Correlation Analysis: Employed in CC-PROMISE for integrating multi-omics data.
  • Bioinformatics Tools and Platforms:
    • Genomic Organismal Subtractive Hybridization (GOSH): A custom software for in silico subtractive genome hybridization.
  • Laboratory Information Management System (LIMS): Developed for biorepository management.
  • Python programming: Used for custom data visualization and analysis.
  • Molecular Evolution and Phylogenetics:
    • PSI-BLAST: For retrieving homologous sequences.
  • Neighbor-joining phylogenetic trees: Constructed using MEGA6 for tracing molecular evolution.
  • Ontology and Knowledge Representation:
    • Biomedical ontologies: Discussed in the context of clinical and translational informatics.

Datasets

The papers presented utilize a variety of datasets, including:

  • Publicly available genomic and transcriptomic datasets:
    • UCSC Genome Browser data.
  • National Sleep Research Resource (sleepdata.org).
  • Gene Expression Omnibus (GEO) data for yeast gene co-expression.
  • The Cancer Genome Atlas (TCGA) data.
  • Human genome browser data.
  • The International HapMap project data.
  • Genome in a Bottle Consortium data.
  • Illumina Platinum Genome call set.
  • Clinical and patient data:
    • Vanderbilt University's BioVU dataset (containing over 215,000 subjects with DNA).
  • Veterans Administration (VA) data for chronic kidney disease (CKD) research.
  • Data from clinical trials (e.g., AML02 clinical trial).
  • Physiological data (e.g., SpO2 levels).
  • Biorepository data.
  • Newly generated or curated datasets:
    • RNA-Seq data from plants infected with coffee ringspot virus (CoRSV).
  • Whole genome shotgun sequencing data from 16 different animals spanning 10 Cetartiodactyla species.
  • RNA-Seq data from three Acinetobacter baumanii clinical isolates.
  • Equine reference genome assembly (EquCab3).
  • Cancer driver gene sequences.

Key Insights and Ideas

  • The power of integrating diverse data types: Several studies emphasize that combining genomic, transcriptomic, epigenomic, and clinical data is crucial for a comprehensive understanding of biological systems and disease mechanisms. This multi-omics approach can reveal insights missed by analyzing single data types alone.
  • Evolutionary history is key to understanding protein function and disease: Understanding how genes and proteins have evolved can provide critical clues about their functional roles and how mutations might lead to disease. This perspective is vital for classifying disease-causing variants.
  • Ontologies are essential for organizing and querying biological knowledge: The use of biomedical ontologies is highlighted as a fundamental step in building robust informatics infrastructure for clinical and translational research, enabling better data sharing and analysis.
  • Computational tools can significantly accelerate biological discovery: The development of new algorithms and software, such as GOSH for genome comparison or NOODLE for integrating miRNA data, demonstrates how computational approaches can overcome limitations in experimental methods and speed up research.
  • The importance of robust benchmarking and validation: Several papers focus on evaluating and comparing different bioinformatics tools and pipelines, underscoring the need for rigorous assessment to ensure reliable results, especially when dealing with clinical data.
  • The potential of deep learning for complex biological problems: The application of deep learning techniques to cancer classification shows promise in handling high-dimensional omics data and achieving improved predictive performance.

Key Results

  • UCSC Genome Browser workshops: Provided hands-on experience with visualization tools for genomic data, covering annotation tracks, track hubs, and database connections.
  • Ontology presentation: Discussed the role of ontologies in clinical informatics, highlighting resources like the National Sleep Research Resource and approaches for ontology quality assurance.
  • Evolutionary history for protein function: Demonstrated how evolutionary analysis can predict functional changes in proteins and help classify disease mutations, with a computational approach developed to consider changes across gene copies.
  • Gene to phenome cataloging: Introduced PrediXcan, an approach for medical informatics data integration, with preliminary results showing novel gene-phenotype relationships.
  • Epigenetics roadmap: Presented on the Roadmap Epigenomics Project, discussing epigenetic markers and tools for accessing this data, with applications in understanding genetic variants and disease.
  • Modeling clinical trials with observational methods: Showcased how Big Data can supplement clinical trials, particularly for chronic kidney disease (CKD) research using Veterans Administration data.
  • Poster session highlights: Featured 24 poster abstracts across various categories, with eight selected for short presentations, covering topics from protein structure analysis to miRNA editing and UTR diversity.
  • NOODLE for DE gene identification: Demonstrated improved differential expression gene identification by integrating miRNA-mRNA interaction networks with microarray data, showing up to a four-fold increase in experimental power.
  • GOSH for genome comparison: Successfully used to identify unique genetic sequences by performing in silico subtractive genome hybridization.
  • PAX gene evolution: Identified conserved domains (Helix-Turn-Helix) in PAX genes implicated in cancer cell survival, suggesting potential drug targets.
  • RNA-Seq data processing: Developed methods for processing RNA-Seq data from virus-infected plants without a reference genome, including quality control, alignment, and transcriptome assembly.
  • Comparative transcriptomics of A. baumanii: Identified transcripts for genes associated with antibiotic resistance mechanisms, suggesting potential biomarkers for rapid diagnosis.
  • Metagenomic assessment of equine genome: Identified potential microbial contaminants in the equine reference genome assembly using Kraken and BLASTX.
  • Biorepository LIMS: Developed a flexible and cost-effective LIMS for biorepository management, integrating EMR data and sample tracking.

Limitations and Future Work

  • Limitations of computational tools: The paper on metagenomic assessment noted that Kraken alone was not sufficient to accurately identify bacterial contaminants and suggested further corroboration with other analytical methods.
  • Challenges in data integration: The paper on personalized oncology highlighted challenges in routine implementation of RNA-Seq profiling, including the lack of normal tissue from the same site for comparison, reference sample size, and appropriate normalization methods.
  • Further characterization of findings: The GOSH paper mentioned ongoing efforts to characterize potential issues with iterative subtractions and repeats found in fungal genomes.
  • Improving motif discovery: The paper on motif discovery using K-means clustering suggested that sensitivity could be improved by allowing mismatches in pattern search.
  • Future directions in collaboration: The summit organizers plan to focus more on bioinformatics research in Kentucky and Tennessee in future summits to foster collaborative efforts.
  • Development of new algorithms and tools: Several papers presented novel computational approaches and tools, indicating a continuous effort to advance the field.
  • Translational research: The overarching goal of many studies is to translate computational findings into clinical applications, drug discovery, and improved patient care.

Summary updated

BibTeX
@article{rouchka2016proceedings,
  author = {Eric C. Rouchka and Julia H. Chariker and Benjamin J. Harrison and Juw Won Park and Xueyuan Cao and Stanley Pounds and Susana C. Raimondi and James R. Downing and Raul C. Ribeiro and Jeffrey Rubnitz and Jatinder Lamba and Bernie J. Daigle Jr. and Deborah Burgess and Stephanie Gehrlich and John C. Carmen and Nicholas Johnson and Chandrakanth Emani and Kalpani De Silva and Michael P. Heaton and Theodore S. Kalbfleisch and Teeradache Viangteeravat and Rahul Mudunuri and Oluwaseun Ajayi and Fatih Sen and Eunice Y. Huang and Mohammad Mohebbi and Luaire Florian and Douglas J. Jackson and John F. Naber and Akm Sabbir and Sally R. Ellingson and Yuping Lu and Charles A. Phillips and Michael A. Langston and Rahul Kumar Sevakula and Raghuveer Thirukovalluru and Nishchal K. Verma and Yan Cui and Mohammed Sayed and Jing Wang and Qi Liu and Yu Shyr and Xiaofei Zhang and Naresh Prodduturi and Gavin R. Oliver and Diane Grill and Jie Na and Jeanette Eckel-Passow and Eric W. Klee and Michael M. Goodin and Mark L. Farman and Harrison Inocencio and Chanyong Jang and Jerzy W. Jaromczyk and Neil Moore and Kelly L. Sovacool and Leon Dent and Mike Izban and Sammed N. Mandape and Shruti S. Sakhare and Siddharth Pratap and Dana Marshall and M. Scotty Depriest and James N. MacLeod and Hanady Adam and Ethan Blandford and Joel Campbell and Joshua Castlen and Brittany Dixon and Ginger Gilbert and Aaron Hall and Philip Kreisle and Jessica Lasher and Bethany Oakes and Allison Speer and Maximilian Valentine and Naga Satya Venkateswara Ra Nagisetty and Rony Jose and Robert W. Rooney and David Hains},
  title = {{Proceedings of the 15th Annual UT-KBRIN Bioinformatics Summit 2016: Cadiz, KY, USA. 8-10 April 2016}},
  journal = {BMC Bioinformatics},
  year = {2016},
  doi = {10.1186/s12859-016-1154-y},
}
Figure