Research Areas for Bioinformatics Methods
Genome-wide Association Interaction Networks
Our network hypothesis of genome-wide associations studies (GWAS) is that most of the information about disease susceptibility is not localized in one gene’s main effect. Rather, this information is distributed throughout a disease-specific network consisting of numerous small gene interactions and other small genetic effects. The network-distributed information about disease risk is the motivation for our Genetic Association Interaction Network (GAIN) and SNPrank approaches. The SNPrank algorithm can be conceptualized as a gene-importance seeker that surfs through the GAIN accumulating bits of information from each genetic node about association with the phenotype. SNPrank represents a modified eigenvector centrality algorithm that ranks the importance of each SNP (single nucleotide polymorphism) due to the network of complex interactions and main effects encoded in the GAIN.
Publications
N. A. Davis, J. E. Crowe, Jr, N. M. Pajewski, B. A. McKinney, “Surfing
a genetic association interaction network to identify modulators of
antibody response to smallpox vaccine,” Genes and Immunity (Nature
Publishing). doi: 10.1038/gene.2010.3; 2010. (open access)
B.A. McKinney, J.E. Crowe, Jr., J. Guo, and D. Tian, “Capturing the
spectrum of interaction effects in genetic association studies by
simulated evaporative cooling network analysis,” PLoS Genetics. 5(3):
e1000432. doi:10.1371/journal.pgen.1000432; 2009. (open access)
N. A. Davis, Ahwan Pandey, and B. A. McKinney. “Real-world comparison
of CPU and GPU implementations of SNPrank: a network analysis tool for
genome-wide association studies,” Bioinformatics. 27 (2): 284-285;2011 (pdf)
Feature Selection
Feature selection is the determination of which genes, proteins, and/or environmental factors are important to include in a classifier or other model of disease susceptibility or other biological endpoint. We are developing feature selection and classification methods for gene and protein expression, structural and functional MRI’s, and GWAS data. A very popular machine learning method used in all areas of bioinformatics is called Random Forests, but is has very limited ability to identify features that are important due to interaction effects. Thus, we developed a method called Evaporative Cooling (EC), which combines Random Forests and Relief-F. EC is able to identify both main and interaction effects.
Publications
B.A. McKinney, J.E. Crowe, Jr., J. Guo, and D. Tian, “Capturing the
spectrum of interaction effects in genetic association studies by
simulated evaporative cooling network analysis,” PLoS Genetics. 5(3):
e1000432. doi:10.1371/journal.pgen.1000432; 2009. (open access)
B.A. McKinney, D. M. Reif, B. C. White, J. E. Crowe Jr., J. H. Moore.
“Evaporative cooling feature selection for genotypic data involving
interactions,” Bioinformatics. 23:2113-2120; 2007. (pdf)
B.A. McKinney, D.M. Reif, M.T. Rock, K. M. Edwards, S. F. Kingsmore,
J.H. Moore, and J.E. Crowe, “Cytokine expression patterns associated
with systemic adverse events following smallpox immunization,” Journal
of Infectious Diseases. 194(4): 36092; 2006. (free pubmed central)
B.A. McKinney, D.M. Reif, M.D. Ritchie and J.H. Moore, “Machine
learning for detecting gene-gene interactions,” Applied Bioinformatics,
5(2):77-88; 2006. (pdf)
D. M. Reif, A.A. Motsinger, B. A. McKinney, J. E. Crowe, Jr., J.H.
Moore, “Integrated analysis of genetic and proteomic data identifies
biomarkers associated with adverse events following smallpox
vaccination,”Genes and Immunity. 10:112-119; 2009. (abstract)
D. M. Reif, A. A. Motsinger, B.A. McKinney, and J. H. Moore, “Feature
selection using a random forest classifier for the integrated analysis
of multiple data types,” Proceedings of the IEEE Symposium on
Computational Intelligence in Bioinformatics and Computational Biology.
pp. 171-178; 2006. (pdf)
Mathematical Modeling of Time-Series Data
At the functional level, biological systems are complex dynamic systems that must balance feedback from intrinsic noise in gene regulatory network and environmental noise from surroundings. Thus, in order to understand how biological systems function properly and why they fail, one needs predictive mathematical models of the dynamic system. One approach we have taken is to pose the problem in terms of nonlinear system identification. We developed a hybrid approach to automatically discover the network topology and parameters of coupled differential equation systems that describes the kinetics of a system of interacting biomolecules. This method combines a grammar-based evolutionary algorithm with an Unscented Kalman particle filter. We are also investigating agent based models.
Publications
B.A. McKinney, “New Informatics approaches for identifying biologic
relationships in time series data,” Wiley Interdisciplinary Reviews:
Nanomedicine and Nanobiotechnology. 1:60-68; 2009. (pdf)
B.A. McKinney and D. Tian, “Grammatical Immune System Evolution for
Reverse Engineering Nonlinear Dynamic Bayesian Models,” Cancer
Informatics. 6:433-447; 2008. (open access)
B.A. McKinney, J.E. Crowe, H.U. Voss, P.S. Crooke, N.L. Barney, and
J.H. Moore, “Hybrid Grammar-based Approach to Nonlinear Dynamical
System Identification from Biological Time Series,” Physical Review E,
Statistical, Nonlinear, and Soft Matter Physics 73, 021912; 2006. (pubmed)
Integrating Data for Computational Antibody-Antigen Docking
Publications
B.A. McKinney, N. Kallewaard, J.E. Crowe, Jr., and J. Meiler, “Using
the natural evolution of a rotavirus-specific human monoclonal antibody
to predict the complex topography of a viral antigenic site,” Immunome
Research. 3:8; 2007. (open access)
N. Kallewaard, B.A. McKinney, Y. Gu, A. Chen, B. V. V. Prasad, and J.E.
Crowe, Jr., “Functional maturation of the human antibody response to
rotavirus,” The Journal of Immunology. 180:3980-3989; 2008. (pdf)
F. Bibollet-Ruche, B.A. McKinney, F.H. Wagner, A. Duverger, A.A.
Ansari, O. Kutsch “Antibody-mediated activation of chimpanzee T cells
via the TCR/CD3 pathway is a function of the anti-CD3 antibody
isotype,” Journal of Virology, 82:10271-10278; 2008. (abstract)
And much more.