goodturing_probs: Computing Good Turing probabilities of encountering gene...

In c7rishi/variantprobs: Computing Probabilities of Gene Variants

Description

Computing Good Turing probabilities of encountering gene variants (including hitherto unobserved variants) based on training gene mutation frequencies

Usage

goodturing_probs(
  counts = NULL,
  r = NULL,
  N_r = NULL,
  m = NULL,
  conf = 1.96,
  N0min = 0,
  N0 = NULL,
  N12_imp = 1,
  N = NULL
)

Arguments

counts	vector of counts or frequencies of the observed variants.
r	unique frequencies.
N_r	frequency of frequency r.
m	training cohort size.
N0min	the minimum value of N0, if known, to be used while estimating N0. Ignored if N0 is not NULL.
N0	the total number of unobserved variants. If NULL, N0 is estimated using Chao's formula.
N12_imp	imputed value of N1 and N2 if either of them is 0. Defaults to 1.
N	The total number of variants, which is sum(Nr) for r >= 0. Used for computation of N0. Ignored if N0 is provided.

Examples

## Not run: 
# load tcga data
data("tcga")
tcga <- data.table::setDT(tcga)

# calculate variant frequencies for KRAS
var_freq <- tcga[Hugo_Symbol == "KRAS",
                 .(v_f = length(unique(patient_id))),
                 by = .(Hugo_Symbol, Variant)
                 ]
v_f <- var_freq$v_f
names(v_f) <- var_freq$Variant

# calculate cohort size
m <- length(unique(tcga$patient_id))


# Good Turing estimates
goodturing_probs(counts = v_f, m = m)

## End(Not run)

c7rishi/variantprobs documentation built on June 23, 2020, 7:42 a.m.