auto_predict_grid: Automatic filtering of signatures for exposure prediction...

View source: R/discovery_prediction.R

auto_predict_gridR Documentation

Automatic filtering of signatures for exposure prediction gridded across specific annotation

Description

Automatic filtering of signatures for exposure prediction gridded across specific annotation

Usage

auto_predict_grid(
  musica,
  table_name,
  signature_res,
  algorithm,
  sample_annotation = NULL,
  min_exists = 0.05,
  proportion_samples = 0.25,
  rare_exposure = 0.4,
  verbose = TRUE,
  combine_res = TRUE
)

Arguments

musica

Input samples to predit signature weights

table_name

Name of table used for posterior prediction (e.g. SBS96)

signature_res

Signatures to automatically subset from for prediction

algorithm

Algorithm to use for prediction. Choose from "lda_posterior", and decompTumor2Sig

sample_annotation

Annotation to grid across, if none given, prediction subsetting on all samples together

min_exists

Threshold to consider a signature active in a sample

proportion_samples

Threshold of samples to consider a signature active in the cohort

rare_exposure

A sample will be considered active in the cohort if at least one sample has more than this threshold proportion

verbose

Print current annotation value being predicted on

combine_res

Automatically combines a list of annotation results into a single result object with zero exposure values for signatures not found in a given annotation's set of samples

Value

A list of results, one per unique annotation value, if no annotation value is given, returns a single result for all samples, or combines into a single result if combines_res = TRUE

Examples

data(musica_annot)
data(cosmic_v2_sigs)
auto_predict_grid(musica = musica_annot, table_name = "SBS96",
signature_res = cosmic_v2_sigs, algorithm = "lda",
sample_annotation = "Tumor_Subtypes")
auto_predict_grid(musica_annot, "SBS96", cosmic_v2_sigs, "lda")

campbio/musicatk documentation built on July 14, 2024, 8:28 a.m.