NCIS (Normalization for ChIP-Seq) estimates normalizing factor between

a ChIP sample and a control/input sample

We support three main data formats:

1. AlignedRead from Bioconductor ShortRead package (with support of commonly used formats, including Eland, MAQ, Bowtie, SOAP and BAM)

2. Bed format, with at least first 6 fields (chrom, start, end, name, score and strand), http://genome.ucsc.edu/FAQ/FAQformat.html#format1

3. MCS (Minimum ChIP-Seq) format, which is similar to and can be easily converted from bed file format. The format is described as below

MCS data should be a data.frame with fields: chr (factor), pos (integer) and strand (factor, "+" and "-")

pos is 5' location; this is different from eland default which use 3' location for reverse strand.

An example Rdata can be found at http://pages.cs.wisc.edu/~kliang/NCIS/example.Rdata

Proper pre-filtering of data is required: reads should be uniquely mapped only; uninformative reads should be filtered, for example, reads mapped to mitochondrial DNA, or Y chromosome for female samples, etc.

Usage: 

#first load all necessary functions

source("http://pages.cs.wisc.edu/~kliang/NCIS/NCIS.R")

#example R code for MCS data format

res <- NCIS(chip.data, input.data, data.type="MCS")

res

#res$est has the estimated normalizing factor

#res$binsize.est has the estimated binsize

#res$r.seq.depth has the normalizing factor computed from sequencing depth ratio

#res$pi0 has the estimated proportion of background reads among ChIP sample

#for data of AlignedRead format

library(ShortRead)

res <- NCIS(chip.data, input.data, data.type="AlignedRead")