In this talk I will provide an evolution of the statistical methods developed in the beginning and throughout the 20th century as used today in the genomics era. The availability of fast computers makes the statistical methods easy to program and to analyze the data using non-open source software such as SAS, STATA, and SPSS among others, and open source software such as R available in CRAN. I will start with R.A. Fisher’s 1918 seminal paper about the small contribution of a gene in traits or phenotypes to the current polygenic risk model, C.R. Henderson’s 1975 paper in best linear unbiased estimation (BLUE) and best linear unbiased predictor (BLUP) and their use in animal and human models, N.E. Breslow and D.G. Clayton’s 1993 paper in generalized linear mixed models (GLMM) and its current application to identify variants or gene in sequencing data, D.R. Cox’s 1972 seminal paper about regression models and life tables and its current application in time to event model to identify association between genetic variants and risk factors-related outcomes or diseases during treatment as well as its application using family data, T. Haavelmo’s 1943 paper in system of simultaneous equations turned to structural equation models (SEM) applied in social sciences and recently for multiple ‘omics outcomes. Finally I will present machine learning techniques and their application in statistics and genomics. I will also provide applications using simulated and real data sets for the majority of topics described above.
