normalize_screens: Normalizes reads for given screens

In HenryWard/orthrus: Scoring Combinatorial CRISPR Screens

Normalizes reads for given screens

Description

Log2 and depth-normalizes reads between a given list of columns and a given column of an earlier timepoint (e.g. T0) specified in the "normalize_name" entry of each screen. Screens with NULL for their "normalize_name" entry are log2 and depth-normalized, but not normalized to earlier timepoints. If a screen to normalize against has multiple replicates, those replicates are averaged before normalization. Multiple replicates for a screen being normalized, however, are normalized separately against the provided early timepoint.

Usage

normalize_screens(
  df,
  screens,
  filter_names = NULL,
  scaling_factor = 1e+06,
  pseudocount = 1,
  min_reads = 30,
  max_reads = 10000
)

Arguments

df	Reads dataframe.
screens	List of screens generated with add_screens.
filter_names	List of screen names to filter based on read counts by.
scaling_factor	Scaling factor for normalizing reads, helps make different replicates comparable but the specific choice typically has little to no impact on results (default 1e6).
pseudocount	Pseudocount for normalized reads, where lower values are such as 1-10 are recommended for most combinatorial screens with sufficient coverage, e.g. 200x and upwards (default 1).
min_reads	Minimum number of reads to keep (default 30, anything below this value will be filtered out).
max_reads	Maximum number of reads to keep (default 10000, anything above this value will be filtered out).

Value

Normalized dataframe.

HenryWard/orthrus documentation built on June 2, 2023, 10:28 p.m.