A data-driven approach to preprocessing Illumina 450K methylation array data

BMC Genomics

Table 1 Table summarising nomenclature of preprocessing methods

	Background adjustment	Between-array normalization	Dye bias correction
naten	n	t	n
nanet	n	n	t
nanes	n	n	s
danes	d	n	s
danet	d	n	t
danen	d	n	n
daten1	d	t	n
daten2	d	t	n
nasen	n	s	n
dasen	d	s	n

We have used a naming convention that encodes key aspects of each method. Vowels were added between the letters for ease of pronunciation. The 1st letter (column 1) indicates whether background adjustment was performed (i.e. the offset between Type I and II probe intensities added to Type I intensities), with ‘d’ indicating background adjustment and ‘n’ indicating no background adjustment. Note daten2 is identical to daten1 but with the addition of a linear model of Sentrix position to obtain smooth background offsets. The 3rd letter (column 2) specifies whether between-array normalization was performed (i.e. between-sample quantile normalization of M and U separately), with ‘s’ indicating between-array normalization applied to Type I and Type II probes separately, ‘t’ indicating between-array normalization applied to Type I and Type II probes together and ‘n’ indicating no between-array normalization. The 5th letter (column 3) specifies whether the dye-bias correction was performed (i.e. quantile normalization of M against U), with ‘s’ indicating dye bias correction applied to Type I and Type II probes separately, ‘t’ indicating dye bias correction applied to Type I and Type II probes together and ‘n’ indicating no dye bias correction.

ISSN: 1471-2164