estimate.rates: Wrapper function to estimate gene- and time specific...

Description Usage Arguments Value Author(s) Examples

Description

Wrapper function to estimate gene- and time specific concentrations for labeled (alpha) and unlabeled (beta) RNA for whole count table The initial values for alpha and beta can be provided, but are otherwise estimated from the data. In case the estimation yields negative or too small values they are replaced by alphabeta.min. For conditions, where no counts are available in both Labeled and Total samples, no rate estimation takes place and NA values are returned.

Usage

1
2
3
4
5
estimate.rates(counts, dispersion, lengths, conditions, conditions.labeling,
  replicates, cross.cont, seq.depths, lab.time, N = 1,
  consider.replicates = TRUE, rep = 1, logloglik = TRUE,
  alpha.initial = NULL, beta.initial = NULL, alphabeta.min = 1e-04,
  gene.indices = NULL, debug = FALSE)

Arguments

counts

count table, counts per gene (rows) and sample (columns)

dispersion

matrix or dataframe with gene-specific disperion estimates (alpha) for labeled and total sample to be used in Negative Binomial. Rownames should correspond to rownames in count matrix

lengths

gene lengths, should be in same order as rownames of count table

conditions

vector of time points/treatment of experiment for each sample

conditions.labeling

vector indicating for each sample if it was labeled RNA or total RNA

replicates

replicate number of each sample

cross.cont

vector with cross contamination rate values for each sample (total RNAseq cross.cont=1)

seq.depths

vector with sequencing depth values for each sample

lab.time

vector of same length as unique conditions, indicating labeling duration

N

number of cells

consider.replicates

should - if available - be replicates taken into account or not? If not, rep replicates is only considered for estimation

rep

if no replicate is available this is replicate that should be chosen

logloglik

should log-loglikelihood function be used?

alpha.initial

initial value for alpha, can be numeric (same for all genes and timepoints), a vector (same for all timepoints of one gene), or a matrix (individual for each gene and time point)

beta.initial

initial value for beta, can be numeric (same for all genes and timepoints), a vector (same for all timepoints of one gene), or a matrix (individual for each gene and time point)

alphabeta.min

minimum value as initial value for alpha and beta rates

gene.indices

indices of genes that should be used for calculation (in order to run code on smaller subset)

debug

should debugging modus be used?

Value

estimates for alpha and beta, expected counts in Labeled and Total and loss for each gene and each labeling time point

Author(s)

Carina Demel

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
data(gene.counts)
lengths = rnbinom(nrow(gene.counts), mu=1000, size=10)
data(spikein.counts)
data(spikein.labeling)
data(spikein.lengths)
spikeins = rownames(spikein.counts)
data(samples)
norm.res = spikein.normalization(spikein.counts, spikeins, spikein.lengths, spikein.labeling, 
	samples=colnames(spikein.counts), samples$conditions.labeling)
seq.depths = norm.res$sequencing.depth
cross.cont = norm.res$cross.contamination
dispersion = estimateGeneDispersion(gene.counts, samples$conditions.labeling, samples$conditions)
fittingres = estimateAlphaBeta.genes(1:12, gene.counts, dispersion, lengths, 
	samples$conditions, samples$conditions.labeling, samples$replicates, cross.cont, seq.depths,
 N=1, consider.replicates=TRUE, lab.time=c(5,5), alpha.initial=1, beta.initial=1)

carinademel/RNAlife documentation built on May 13, 2019, 12:43 p.m.