calc_norm_factors: calc_norm_factors

In jrboyd/seqsetvis: Set Based Visualizations for Next-Gen Sequencing Data

calc_norm_factors

Description

Calculate normalization factors in a two step process:

Usage

calc_norm_factors(
  full_dt,
  value_ = "y",
  cap_value_ = "y_cap_value",
  by1 = "id",
  by2 = "sample",
  aggFUN1 = max,
  aggFUN2 = function(x) quantile(x, 0.95)
)

Arguments

full_dt	a data.table, as returned by ssvFetch*(..., return_data.table. = TRUE)
value_	character, attribute in full_dt to normalzie.
cap_value_	character, new attribute name specifying values to cap to.
by1	character vector, specifies attributes relevant to step 1.
by2	character vector, specifies attributes relevant to step 1 and 2.
aggFUN1	function called on value_ with by = c(by1, by2) in step 1.
aggFUN2	function called on result of aggFUN1 with by = by2 in step 2.

Details

1. summarize every region for each sample (default summary function is max)

2. caclulate a value to cap each sample to based on regions (default is 95th quantile).

The uderlying assumption here is that meaningful enrichment is present at the majority of regions provided. If prevalence varies by a specific factor, say ChIP-seq targets with different characteristics - ie. when analyzing TSSes for H3K4me3 and an infrequent transcription factor it is more appropriate to specify appropriate quantile cutoffs per factor.

Value

data.table mapping by2 to cap_value_.

Examples

data(CTCF_in_10a_profiles_dt)
calc_norm_factors(CTCF_in_10a_profiles_dt)
calc_norm_factors(CTCF_in_10a_profiles_dt,
  aggFUN1 = mean, aggFUN2 = function(x)quantile(x, .5))

jrboyd/seqsetvis documentation built on Dec. 10, 2024, 11:23 a.m.