generate_model: Generate Model

In andygxzeng/ECSI: Espresso - Somatic Mutation Calling using Contextual Error Models

Description

Generate all error models for each trinucleotide variant context. Fits error distribution with either an Exp or Weibull distribution depending on the overall distribution of non-reference alleles. If the most frequent non-reference allele count is 1 (typically at <10,000 sequencing depth) an Exponential distribution will be fitted and if it is greater than 1 (seen at ultra-deep read depths) a Weibull distribution will be fitted.

Usage

generate_model(i, data, groups, model = "auto")

Arguments

i	Index of the FlankingSeqGroup for which the model is being generated
data	Dataframe from generate_all_models with FlankingSeqGroup and three empty columns for parameters of model fit
groups	Dataframe with 192 rows and 4 columns. First column contains the 192 trinucleotide contexts/FlankingSeqGroup
model	Specifying which error model (exp or weibull) to fit. Default is "auto".

Value

This function returns a dataframe with the following information:

• FlankingSeqGroup

• Model Used (Exp or Weibull)

• Model parameter 1

• Model parameter 2

andygxzeng/ECSI documentation built on Feb. 6, 2021, 8:53 a.m.