IDR.3component: Differential expression analysis by integrating data from...

In Yafei611/idr3c: Differential gene expression analysis by integrating data from multiple sources

Description

Mapping genes expression data from different sources to a normal mixture. Then performing model based cluster and computing irreproducibility rate (idr) of being differential expressed for each genes.

Usage

IDR.3component(x, para = list(p = c(0.4, 0.3, 0.3), mu = cbind(c(0, 0), c(3,
  3), c(-3, -3)), sd = cbind(c(1, 1), c(1, 1), c(1, 1)), rho = c(0, 0.84,
  0.84)), f_ctrl = list(max.t.inner = 10000, max.t.outer = 10000, b.em =
  1e-06, b.outer = 0.001), disp = list(z.rankplot = FALSE, z.rankplot.cut =
  0.2, outer.trace = TRUE, inner.trace = FALSE))

Arguments

x	A matrix or data frame of log-fold change of normalized gene expression data, i.e., of non-negative values. The rows correspond to observations (e.g., expression level for a gene), the columns correspond to data sources (labs/platforms etc.). Note that technical replicates (i.e., several sequencing ruins/lanes from the same sample) should be took the average.
para	A list that contains the parameter of the normal mixture where the pseudo data were generated. $p The size of each components (proportion of non/up/down regulated genes) $mu Means for each normal component. para$mu should be a 2-by-3 matrix, each row are means for the same components. $sd Standard deviations for each normal component. para$sd should be a 2-by-3 matrix, each row are standard deviations for the same components. $rho Pearson correlation for each normal component.
f_ctrl	A list that contains the parameter controls pseudo-EM procedure. The default f_ctrl usually works well. $max.t.inner The maximum iterations numbers of the inner EM algorithm $max.t.outer The maximum number of times that re-generating pseudo data $b.em Stop criterion for inner EM algorithm. $b.outer Stop criterion for outer iteration.
disp	A list control the display of the trace of parameter estimation. $z.rankplot Logical. Should the plot of pseudo data be displayed when updated? $z.rankplot.cut The proportion of samples with low idr that will be marked on the plot. $outer.trace Logical. Should the trace of the parameter estimation be displayed? $inner.trace Logical. Should the trace of the parameter estimation for the inner EM be displayed?

Details

Please refer Lyu, Y., & Li, Q. (2016). A semi-parametric statistical model for integrating gene expression profiles across different platforms. BMC Bioinformatics, 17(1), 51.

Value

IDR.3component produces a list that contain idr value, parameter estimation and other necessary components.

• $para The final estimation of the normal mixture

• $loglik.trace.in Trace of log-likelihood of EM produce

• $loglik.trace.ot Trace of log-likelihood of outer iterations

• $idr the posterior probability that a gene is in the non-DEGs group

• $IDR computed irreproducibility rate

Examples

x = expr
plot(x[,1],x[,2],xlim=c(-8,8),ylim=c(-8,8),cex = .4);
idr.out = IDR.3component(x = x)

Yafei611/idr3c documentation built on May 9, 2019, 11:07 p.m.