Code
Show All Code
Hide All Code
Predict Variant Diagnosis Probability Using Structure, Function, and
In Silico
Features (
KCNH2
and LQT2)
Brett Kroncke
October 01, 2020
Introduction
Part 1: Calculate probability of LQT2 diagnosis and LQT2 Probability Density using Various Subsets of the Literature and Cohort Data
All Literature Variants
LQT2 empirical LQTS diagnosis probability prior
Calculate LQTS probability densities
Calculate Weighted Spearman Correlation Coefficients
Scale all covariates
Calculate EM priors and posteriors for all variants
Literature Variants Where N = 1 Variants are removed
LQT2 empirical diagnosis probability prior
Calculate LQTS probability densities
Scale all covariates
Calculate EM priors and posteriors for all variants
Literature and Cohort Combined (for final predictions)
LQT2 empirical diagnosis probability prior
Calculate LQTS probability densities and annotate function and structural location
Calculate EM priors and posteriors for all variants
LQT2 diagnosis probability in Cohort Data for validation
Part 2: Coverage plots
Bootstrap and get the coverage rate
Bootstrap function
Plot coverage
Observed diagnosis probability as the “true” diagnosis probability
Part 3: Variance explained
Pearson R^2 and Spearman Rho Against EM Posterior from Cohort
Variance explained from literature dataset
Part 4: ROC and AUC plots