Bioinformatics Core

点击翻译按钮获取中文摘要

Abstract The Bioinformatics Core (CORE B) will interface with each projects, cores, and UTSW consortium to provide (i) centralized bioinformatics and biostatistical support; (ii) centralized database; (iii) integrative data analysis across different platforms; and (iv) analytical and methodical reports for preparation of manuscripts. Core B Leader, Dr. Wang, and Co-Leader, Dr. Baladandayuthapani, have been working closely and synergistically for several years with core and projects leaders and co-leaders, and are capable of supporting study designs, data analysis, and management of data resources of the entire PDTC program. Core B will use robust IT structure and an extensive computing environment that includes Windows, Unix/Linux, Mac OS X, and two quad-processor Sun SPARC SMP systems. In addition, Core B will rely on two primary institutional computing resources, an HPC cluster of 336 compute nodes (each node with 32GB RAM; 1.3 GB RAM per core) with dual, 12-core Opteron processors (8,064 CPUs total), and an Itanium-2 SMP compute server with 32CPUs and 128GB RAM. The core Leader and Co-Leader have built various pipelines for sequencing data and protein expression data processing and analysis, which will be applied to analyze data. Standard design principles and statistical algorithms will be used and new methods will be developed as needed. These include: a) Parametric and nonparametric methods for estimation and hypothesis testing; b) Linear models and generalized additive models to find the best models that fit complex data structures; c) Kaplan-Meier method to estimate the distributions of time-to-event outcomes; d) Log-rank test to compare the distributions among different PDX groups; and e) Proportional hazards models to test for PDXs treatment with single drugs and combinations. Drs. Wang and Baladandayuthapani will work closely with UTPDTC investigators to facilitate hypothesis testing across projects by integrating datasets from multiple laboratories using various algorithms, including principal components, partial least squares, and Bayesian network-based models. Data analyses will be performed using R and Bioconductor packages. The Core will document all the analyses and produce HTML or PDF reports (using R packages: Sweave, knitR, and R markdown).