CHSI’s statistical/computational immunology division carries out statistical/computational design and analysis for our collaborative projects with investigators. These projects range from correlative immune studies to building predictive models for disease risk. Our team also develops novel statistical methods for research questions that cannot be addressed by existing quantitative analyses.
- Exploration, analysis and visualization of correlative immune studies. The statistical inference group can provide support for experimental design and analysis of correlative studies and biomarker discovery from soluble factor, antibody, cellular, transcriptomic, or imaging studies.
- Development of custom statistical models for immune-related assays. Existing analytic methods for specific assays may be inadequate or inappropriate for the needs of the investigator. The CHSI statistical inference group can help develop new statistical methods to rigorously analyze assay data. The development of new methods will be led by a faculty member working with a graduate student or postdoc. CHSI will provide partial funding (50%) for a graduate student/postdoc for the development of new methods likely to be of high impact.
- Integrative analysis of data from multiple assay platforms. We are interested in initiating collaborative projects that involve integrating data from more than one assay platform to provide a more complete picture of the immune system. Please contact us if you have a project that could benefit from such “data fusion” and comes with or will generate such multi-platform data.
- Genomics of highly polymorphic loci. We are interested in the imputation of highly polymorphic genomic loci relevant to immunology such as the HLA, KIR, and FcR regions, and in performing association analysis of the genomic variation with immune-related diseases.
- Polygenic risk score for immune diseases. Constructing a predictive model for the PRS requires jointly analyzing genome-wide SNPs simultaneously on large GWAS datasets. CHSI has developed Bayesian model averaging and Markov Chain Monte Carlo (MCMC) to derive such prediction models.