BeviMed provides functions for performing Bayesian Evaluation of Variant Involvement in Mendelian Disease. The aim is to estimate the probability of an association between a latent configuration of a restricted set of variants and a case-control label, and conditional on the association, the probabilities of pathogenicity for individual variants.
The inference is carried out based on the inputs:
y, a length N (number of samples) logical vector,G, a k by N integer matrix of allele counts,min_ac, representing a mode of inheritance hypothesis (i.e. minimum number of pathogenic variants required to be considered to have a pathogenic configuration of variants).Then, depending on the quantity of interest, the inference procedure can be invoked simply by passing the above arguments to the functions:
prob_association - returning the probability of association between configurations of variants represented in G and the case-control label y (optionally broken down by mode of inheritance),log_BF - the log Bayes factor between the association model and no-association model,prob_pathogenic - the probabilities of pathogenicity for the individual variants.The inference is performed by the function bevimed, an MCMC sampling procedure with many parameters, including those listed above and others determining the sampling management and prior distributions of the model parameters.
It returns a list of traces for the sampled parameters in an object of class BeviMed. This object can take up a lot of memory, so it may be preferable to store a summarised version passed to summary.
Here we demonstrate a simple application of BeviMed for some simulated examples.
library(BeviMed)
set.seed(0)Firstly, we’ll generate some random data consisting of a case-control label, y and a variant-wise indicator of pathogenicity, Z, where Z[i] == TRUE if variant i is pathogenic and FALSE otherwise.
In y, we will represent 100 samples, the first 20 of whom are cases and the next 80 of whom are controls, and in Z we will represent 20 SNPs at which the samples are to be genotyped, of which the first 3 are pathogenic and of which the next 17 are benign.
y <- c(rep(TRUE, 20), rep(FALSE, 80))
Z <- c(rep(TRUE, 3), rep(FALSE, 17))In the first application, we’ll use variant matrix G1 where there is no association between y and the variants (i.e. we simulated all genotypes at random).
G1 <- sapply(y, function(y_i) as.integer(runif(n=length(Z)) < 0.15))
prob_association(G=G1, y=y)## [1] 0.01511634The results indicate that there is a low probability of association (by default a prior probability of 0.01 with each mode of inheritance is used). In the example with association, we’ll sample the allele counts based on y (i.e. for the cases we’ll include a random pathogenic variant amongst the observed genotypes).
G2 <- sapply(y, function(y_i) as.integer(runif(n=length(Z)) < 0.15 |
(if (y_i) 1:length(Z) == sample(which(Z), size=1) else rep(FALSE, length(Z)))))
prob_association(G=G2, y=y)## [1] 0.9987681Notice that there is now a higher estimated probability of association.
By default, prob_association integrates over mode of inheritance (e.g. are at least 1 or 2 pathogenic variants required for a pathogenic configuration?). The probabilities of association with each mode of inheritance can by shown by passing the option by_MOI=TRUE (for more details, including how to set the ploidy of the samples within the region, see ?prob_pathogenic).
For a more detailed output, the bevimed function can be used, and it’s returned values can be summarised and stored/printed.
output <- summary(bevimed(G=G2, y=y))
output## ---------------------------------------------------------------------------
## The probability of association is 1 [prior: 0.01]
##
## Log Bayes factor between gamma 1 model and gamma 0 model is 11.16
##
## A confidence interval for the log Bayes factor is:
## 2.5% 97.5%
## 10.41 11.81
## ---------------------------------------------------------------------------
## Estimated probabilities of pathogenicity of individual variants
## (conditional on gamma = 1)
##
## Variant Controls Cases P(Z_j=1|y,gamma=1) Bar Chart
## 1 9 7 1.00 [=================== ]
## 2 17 9 1.00 [====================]
## 3 10 6 1.00 [====================]
## 4 12 2 0.00 [ ]
## 5 14 2 0.00 [ ]
## 6 11 4 0.01 [ ]
## 7 12 4 0.01 [ ]
## 8 13 2 0.01 [ ]
## 9 11 2 0.00 [ ]
## 10 12 4 0.00 [ ]
## 11 11 4 0.01 [ ]
## 12 10 4 0.00 [ ]
## 13 16 5 0.00 [ ]
## 14 8 4 0.01 [ ]
## 15 16 4 0.00 [ ]
## 16 11 1 0.01 [ ]
## 17 14 2 0.00 [ ]
## 18 14 1 0.01 [ ]
## 19 11 3 0.00 [ ]
## 20 7 2 0.01 [ ]