mh_chain: Metropolis-Hasting for a single peptide

Description Usage Arguments

View source: R/immunogen_mh_tools_2.r

Description

Implementation of the Metropolis-Hastings algorithm to find posterior distributions of pHLA immunogenicity for a single peptide

Usage

1
2
mh_chain(eli.dat, mol.names, init_h, max_steps = 5000, pep, unif.prop = T,
  radius = 0.1, p.df = NULL)

Arguments

eli.dat

ELISpot dataset

mol.names

Names of the HLA molecules in the ELISpot dataset

init_h

Initial state of the MH chain, can be produced randomly by generate_random_hypothesis

max_steps

The required length of the chain. The default value, 5000, is too low but this was done for debugging reasons.

pep

Peptide currently being processed

unif.prop

If TRUE, choose proposal states from [0,1]^m

radius

If unif.prop is FALSE, we look for a proposal state in a hypercube around h restricted by radius. Should be trained such that the acceptance rate of the chain is around 50%, see example.r.

p.df

Dataframe which describes the parameters of the prior distributions for each HLA


liesb/BIITE documentation built on May 21, 2017, 1:35 p.m.