An evaluation of methods correcting for cell-type heterogeneity in DNA methylation studies

Table 11 Comparison of some features of the methods for cell-type mixture adjustment

Method	Phen. allowed ^a	Input values ^b	K ^c	Link
Ref-based	Any	Beta	N/A	http://people.oregonstate.edu/~housemae/software/TutorialLondon2014
				http://bioconductor.org/packages/release/bioc/html/minfi.html
Ref-free	Any	Beta	Estimated	http://cran.r-project.org/web/packages/RefFreeEWAS/index.html
SVA	Any	Beta or logit(beta)	Estimated	http://bioconductor.org/packages/release/bioc/html/sva.html
ISVA	Continuous	Beta or logit(beta)	Estimated	http://cran.r-project.org/web/packages/isva/index.html
EWASher	Binary	Beta	Estimated	http://research.microsoft.com/en-us/downloads/472fe637-7cb9-47d4-a0df-37118760ccd1
CellCDec	Not used	Beta	Input	https://github.com/jameswagner/CellCDec
Deconf	Not used	Beta or logit(beta)	Input	http://web.cbio.uct.ac.za/~renaud/CRAN
RUV	Any	Beta or logit(beta)	Estimated	https://cran.r-project.org/web/packages/ruv/index.html

^aWhat kinds of phenotype are allowed?
^bDoes the method use methylation proportions (beta values)? Or logit transformed beta values?
^cDoes the method estimate the number of latent cell types K, or is K input into the algorithm?

ISSN: 1474-760X