TS_kernel: Calculate the Test Statistics for kernel-based nonparametric...

In ChIPtest: Nonparametric Methods for Identifying Differential Enrichment Regions with ChIP-Seq Data

Description

Get the difference between two conditions. Apply Kernel smoothing to fit a smooth curve. Estimate variance for each gene and improve the estimation of variance based on all the genes. Derive test statistics and get the rank list of all the genes.

Usage

TS_kernel(data, band, quantile)

Arguments

data	difference matrix between two conditions
band	bandwidth used in kernel smoothing
quantile	threshold used in variance estimation

Details

Note 1: Need to chose a bandwidth. Do not recommend to use cross validation (not gene-specific bandwidth) but chose a fixed biological meaningful bandwidth. A fixed bandwidth which can capture the signal profile and smooth out noise would be recommend. The bandwidth used in reference is 20/280.

Note 2: quantile value is based on the distribution of variance estimation of each gene. Recommend to use histogram to double check the distribution. Default 0.9 = 90 %

Value

TS	Kernel based test statistics after WH transformation. Please refer the details in the reference
TS_sign	"+" represent for condition B enriched more than condition A; "-" vice versa
Tmean	Original test statistics, which is calculated as integral of square of kernel estimator

References

Qian Wu, Kyoung-Jae Won and Hongzhe Li. (2015) Nonparametric Methods for Identifying Differential Enrichment Regions with ChIP-seq Data. Cancer Informatics,14 (Suppl 1), 11-22

Examples

data(data1)
data(data4)
Data1=NormTransformation(data1)
Data4=NormTransformation(data4)
data=Data4-Data1
band=54
TS=TS_kernel(data, band, quantile=0.9)

ChIPtest documentation built on May 2, 2019, 3:37 p.m.