guessInits: Derive initial estimates of unknown model parameters

In athchen/beer: Bayesian Enrichment Estimation in R

Derive initial estimates of unknown model parameters

Description

To reduce converge time and to reduce the likelihood of the slice sampler getting stuck, we use maximum likelihood to derive initial estimates for unknown model parameters.

Usage

guessInits(object, beads.prior)

Arguments

object	a PhIPData object
beads.prior	a data frame with two columns (named a_0, b_0) containing estimated shape parameters from beads-only samples.

Details

Briefly initial values are defined as follows:

1. theta_guess[i, j] = Y[i, j]/n[j], or the the MLE for theta.

2. Z_guess[i, j] = 1 if j is a serum sample, and the observed read count is >2x the expected read count assuming c[j] = 1.

3. pi_guess[j] is the mean of column j in Z_guess.

4. c_guess[j] is the estimated slope from regressing the observed read counts against the expected read counts (without adjusting for the attenuation constant) for non-enriched peptides only.

5. phi_guess[i,j] is the ratio of the observed read counts to the expected read counts multiplied by the attenuation constant.

Value

a list of estimated initial values.

See Also

Methods in [Chen et. al 2022](https://www.biorxiv.org/content/10.1101/2022.01.19.476926v1)

athchen/beer documentation built on July 2, 2022, 10:35 p.m.