SMSC: SMSC estimates the parameters of the mixture of binomial...

In KarimOualkacha/SmoothMSC: The smooth methylation status calling (SMSC) procedure

Description

we propose statistical methods for single cell MethylC-Seq and RRBS data that simultaneously take into account the spatial correlation between DNA methylation profiles at neighbouring CpG sites and correct for the experimental and sequencing errors. Following Cheng & Zhu (2014), we model the observed count of C reads in the presence of sequencing errors with a mixture of binomial distributions. In our case, we consider a methylation probability that varies with the genomic location and derive a smoothed kernel-based EM-algorithm to estimate the parameters of our model.

Usage

SMSC(C, CT, t, p1 = 0.85, p2 = 0.1, method = c("Locfit", "KNN1", "KNN2", "MSC"), eps = 1e-06, K = 70, h = 1000, maxIt = 100)

Arguments

C	the observed count of C reads at the ith CpG site
CT	the observed count of C reads and T reads for each site i
t	The location of the site i on a chromosome
p1	The error rate for obtaining C reads at unmethylated sites
p2	the compelement of p2 (1-p2) denote the error rate for obtaining T reads at methylated sites
method	a character string specifying the ckeck the non parametric method to use for smoothing the probability that site i is methylated, valid options are: "Locfit": we use the . "KNN1": we use the . "KNN2": we use the . "MSC": the probability in the site i is the average of all sites (procedure MSC: Cheng et al 2014), Default is "Locfit"
eps	convergence threshold for EM algorithm using smoothing technic
K	the kth nearest neighbor
h	the bandwidth
maxIt	the maximum number of iterations allowed.

Value

An object with S3 class SMSC.

References

• Hansen, K.D., Langmead, B. & Irizarry, R.A. BSmooth: from whole genome bisulfite sequencing reads to differentially methylated regions. Genome Biol. 13, R83 (2012).

• Cheng L, Zhu Y. A classification approach for DNA methylation profiling with bisulfite next-generation sequencing data. Bioinformatics. 2014; 30(2):172–9.

• Rackham OJ, Dellaportas P, Petretto E1, Bottolo L: WGBSSuite: simulating whole-genome bisulphite sequencing data and benchmarking differential DNA methylation analysis tools. 10.1093/bioinformatics/btv114. Epub Mar 15

Parameswaran Ramachandran,and Theodore J Perkins: Adaptive bandwidth kernel density estimation for next-generation sequencing data. BMC Proc. 2013; 7(Suppl 7): S7.

Examples

#### MSC scenario
data(Data_MSC)
#### HHM scenario
data(Data_HHM)

y <- Data_MSC #### or y <- Data_HHM

fit0<-SMSC(y$Ccount, y$CT,y$position, method="MSC")

fit1<-SMSC(y$Ccount, y$CT,y$position, method="KNN1")

fit2<-SMSC(y$Ccount, y$CT,y$position, method="KNN2")

fit3<-SMSC(y$Ccount, y$CT,y$position, method="Locfit")

KarimOualkacha/SmoothMSC documentation built on May 8, 2019, 4:49 p.m.