omniBiomarker Gene Ranking Algorithms
omniBiomarker's gene ranking algorithms include a mixture of common univariate and multivariate filter methods, including
  • Fold Change (FC)
  • T-Test (T)
  • Wilcoxon Rank Sum Test (RS)
  • Significance Analysis of Microarrays (SAM)1
  • Rank Products (RP)2
  • minimum Redundancy Maximum Relevance w/ Difference (MRMRD)3
  • minimum Redundancy Maximum Relevance w/ Quotient (MRMRQ)3
It is important to select a gene ranking method to maximize the biological relevance of ranking.4 We measure the biological relevance of these ranking methods using reference knowledge. Although the ground-truth of gene ranking is generally not known, we can use previously validated differentially expressed genes to guide algorithm selection and overcome the curse-of-dimensionality in biomarker identification problems.5 Currently, we provide a default knowledge set of genes from the Cancer Gene Index project.6 In addition, users can provide their own list of knowledge genes.
Distributed Grid Computing with BOINC
We use the Berkeley Open Infrastructure for Network Computing (BOINC) as the framework for omniBiomarker's high-performance computing. BOINC, which is used for volunteer-based scientific computing projects such as Folding@Home and Genome@Home, is a desirable computing framework for omniBiomarker due to its scalability.7 Although we only use BOINC as a desktop grid computing solution, its flexibility enables us to potentially expand to volunteer-based computing in order to more effectively analyze the increasing amounts of gene expression data, biological knowledge, and biomarker identification algorithms.

caBIG Analytical Grid Services
The accessibility of bioinformatics tools has become increasingly important as the gap between clinical applications and bioinformatics narrows. To improve the accessibility of omniBiomarker, we have developed Cancer Biomedical Informatics Grid (caBIG) Silver Level compliant grid services to access basic gene ranking and data retrieval functions. caBIG certification increases the interoperability of omniBiomarker's functionality with other bioinformatics tools in the cancer research community. Currently, omniBiomarker includes two grid services:
  • GeneRankAnalysis rankGenes(GeneRankParameters)
  • GeneRankResults getGeneRanks(GeneRankAnalysis)
These services are the first step towards full interoperability. The standardized data structures required by these services are illustrated below:

  1. Tusher VG, Tibshirani R and Chu G. Significance analysis of microarrays applied to the ionizing radiation response.. Proc Natl Acad Sci USA. 2001. 98(9):5116-21. pubmed
  2. Breitling R, Herzyk P. Rank-based methods as a non-parametric alternative of the T-statistic for the analysis of biological microarray data. J Bioinform Comput Biol. 2005. 3(5):1171-89. pubmed
  3. Ding C and Peng H. Minimum redundancy feature selection from microarray gene expression data. J Bioinform Comput Biol. 2005. 3(2):185-205. pubmed
  4. Phan JH, Moffitt RA, Stokes TH, Liu J, Young AN, Nie S and Wang MD. Convergence of biomarkers, bioinformatics and nanotechnology for individualized cancer treatment. Trends Biotechnol. 2009. 27(6):350-8. pubmed
  5. Phan JH, Yin-Goen Q, Young AN and Wang MD. Improving the efficiency of biomarker identification using biological knowledge. Pac Symp Biocomput. 2009. 14:427-38. pubmed pdf
  6. Schueller CME, Fritz A, Torres Schumann E, Wenger K, Albermann K, Komatsoulis GA, Covitz PA, Wright LW, and Hartel F. Towards a comprehensive catalog of gene-disease and gene-drug relationships in cancer.. Thirteenth ISMB 2005 International Conference on Intelligent Systems for Molecular Biology, Detroit, MI, USA. 2005.
  7. Larson SM, Snow CD, Shirts M and Pande VS. Folding@Home and Genome@Home: Using distributed computing to tackle previously intractable problems in computational biology. 2009.
Powered By:

Please use Firefox or Google Chrome. Support for Internet Explorer is still under development.