#Copy number estimation from logR using ASCAT formula

Requires logR values and previously defined segments (doesn't do segmentation!)

##Setting up the model

You need to provide:

• a vector of logR values
• number of segments
• indices of start of each segment
• indices of end of each segment
• prior parameters

Build the model and initialise it with data:

jmod <- jags.model("copynumber.bug",
    data = list(
        logr = logr_vector,
        nsegments = number_of_segments,
        starts = left_hand_position_of_breakpoints,
        ends = right_hand_position_of_breakpoints,
        psi_a = 4,    # psi_a, psi_b: parameters
        psi_b = 2,    # of gamma prior on psi (ploidy)
        rho_a = 1,    # rho_a, rho_b: parameters
        rho_b = 1,    # of beta prior on rho (purity)
        sd = 0.1),    # standard dev of normal distribution
    n.chains = 1,   # number of MCMC chains to run
    n.adapt = 100)  # number of cycles to spend tuning the model before running

Running the model

# burn-in 1000 steps
update(jmod, 1000)

# sample 1000 iterations of the MCMC (n.iter / thin = 1000)
jsamples <- jags.samples(
    jmod,
    c("CNplus1", "rho", "psi"),
    n.iter = 1000,
    thin = 1)

Getting the results

CN.est <- summary(jsamples$CNplus1, mean)$stat - 1
rho.est <- summary(jsamples$rho, mean)$stat
psi.est <- summary(jsamples$psi, mean)$stat