SWAN: Subset-quantile within array normalization for illumina infinium HumanMethylation450 BeadChips - PubMed
- ️Sun Jan 01 2012
SWAN: Subset-quantile within array normalization for illumina infinium HumanMethylation450 BeadChips
Jovana Maksimovic et al. Genome Biol. 2012.
Abstract
DNA methylation is the most widely studied epigenetic mark and is known to be essential to normal development and frequently disrupted in disease. The Illumina HumanMethylation450 BeadChip assays the methylation status of CpGs at 485,577 sites across the genome. Here we present Subset-quantile Within Array Normalization (SWAN), a new method that substantially improves the results from this platform by reducing technical variation within and between arrays. SWAN is available in the minfi Bioconductor package.
Figures

Illumina Infinium HumanMethylation450 assay. (a) Infinium I assay. Each individual CpG is interrogated using two bead types: methylated (M) and unmethylated (U). The probe design assumes that all CpGs underlying the probe body have the same methylation status as the target CpG. Both bead types will incorporate the same labeled nucleotide for the same target CpG, thereby producing the same color fluorescence. The nucleotide that is added is determined by the base downstream of the 'C' of the target CpG. The proportion of methylation, β, can be calculated by comparing the intensities from the two different probes in the same color: β= M/(U + M). (b) Infinium II assay. Each target CpG is interrogated using a single bead type. A probe may have up to three underlying CpG sites, with a degenerate R base corresponding to the 'C' of each CpG. Methylation state is detected by single base extension at the position of the 'C' of the target CpG, which always results in the addition of a labeled 'G' or 'A' nucleotide, complementary to either the 'methylated' C or 'unmethylated' T, respectively. Each locus is detected in two colors, and methylation status is determined by comparing the two colors from the one position: β = Green (M)/(Red (U) + Green (M)). (c) The number of CpG dinucleotides in the body of the probe according to Infinium probe type. Infinium I probes have significantly more CpGs in the probe body.

Intensity distributions of the Infinium I and II probe types in the methylated and unmethylated channels for normal human kidney sample TCGA-B0-5092-11. The qualitative differences in the intensity distributions are probably driven by the biological differences between the regions that the two probe types are interrogating, which is reflected by the difference in density of CpGs in the body of the probes.

Intensity distributions of subsets of Infinium I and II probes with the same number of underlying CpGs for normal human kidney sample TCGA-B0-5092-11. (a) One CpG in the probe body. (b) Two CpGs in the probe body. (c) Three CpGs in the probe body.

Differences in β value distributions produced by Infinium I and II probes. (a) Density distributions of β values produced by Infinium I (solid line) and II (dashed line) probes for unmethylated (blue), hemi-methylated (purple) and methylated (red) reference standards (Std.). The difference in β value distribution between the Infinium I and II probes seen in the raw data can be adjusted for using the SWAN method. (b) The median and inter-quartile range of β value distributions for Infinium I and II probes is more similar when SWAN is applied to 450k data. (c) The differences in β value distributions produced by the different probe types can result in aberrant overall β value distributions, as seen in this normal human DNA sample. Applying the SWAN method results in an improvement in the overall β value distribution.

β value density distributions for four pairs of technical replicates before and after applying SWAN. (a-d) β value density distributions for each pair of technical replicates. The Kolmogorov-Smirnov (KS) test P-value reflects the similarity of the β value distributions between each pair of replicates; a larger P-value indicates that the distributions of the replicates are more similar.

Comparison of β value density distributions between HumanMethylation450 and HumanMethylation27 arrays before and after SWAN in an MCF7 cell line. This plot illustrates the β value density distributions for 25,978 CpGs that are present on both 450k and 27k platforms. The peaks of the 450k Infinium I and II probe types show better alignment with the 27k peaks when SWAN is used.

Results of differential methylation analysis of three kidney samples compared to three rectum samples, with and without using SWAN. (a) Percentage of RRBS true positives identified at various q-value significance thresholds. Using the SWAN method (magenta) consistently detects more RRBS true positives than analyzing raw data (black). (b) Receiver operating characteristic (ROC) curve of an analysis using SWAN compared to an analysis of the raw data. Using SWAN (magenta) prior to differential methylation analysis results in performance gains.
Similar articles
-
Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi.
Fortin JP, Triche TJ Jr, Hansen KD. Fortin JP, et al. Bioinformatics. 2017 Feb 15;33(4):558-560. doi: 10.1093/bioinformatics/btw691. Bioinformatics. 2017. PMID: 28035024 Free PMC article.
-
missMethyl: an R package for analyzing data from Illumina's HumanMethylation450 platform.
Phipson B, Maksimovic J, Oshlack A. Phipson B, et al. Bioinformatics. 2016 Jan 15;32(2):286-8. doi: 10.1093/bioinformatics/btv560. Epub 2015 Sep 30. Bioinformatics. 2016. PMID: 26424855
-
Analysis pipelines and packages for Infinium HumanMethylation450 BeadChip (450k) data.
Morris TJ, Beck S. Morris TJ, et al. Methods. 2015 Jan 15;72:3-8. doi: 10.1016/j.ymeth.2014.08.011. Epub 2014 Sep 16. Methods. 2015. PMID: 25233806 Free PMC article. Review.
-
A comprehensive overview of Infinium HumanMethylation450 data processing.
Dedeurwaerder S, Defrance M, Bizet M, Calonne E, Bontempi G, Fuks F. Dedeurwaerder S, et al. Brief Bioinform. 2014 Nov;15(6):929-41. doi: 10.1093/bib/bbt054. Epub 2013 Aug 29. Brief Bioinform. 2014. PMID: 23990268 Free PMC article. Review.
Cited by
-
Koeck RM, Busato F, Tost J, Zandstra H, Remy S, Langie S, Gielen M, van Golde R, Dumoulin JCM, Brunner H, Zamani Esteki M, van Montfoort APA. Koeck RM, et al. Hum Reprod. 2022 Oct 31;37(11):2709-2721. doi: 10.1093/humrep/deac213. Hum Reprod. 2022. PMID: 36206092 Free PMC article.
-
Chen YA, Lemire M, Choufani S, Butcher DT, Grafodatskaya D, Zanke BW, Gallinger S, Hudson TJ, Weksberg R. Chen YA, et al. Epigenetics. 2013 Feb;8(2):203-9. doi: 10.4161/epi.23470. Epub 2013 Jan 11. Epigenetics. 2013. PMID: 23314698 Free PMC article.
-
Lim AM, Candiloro IL, Wong N, Collins M, Do H, Takano EA, Angel C, Young RJ, Corry J, Wiesenfeld D, Kleid S, Sigston E, Lyons B, Rischin D, Solomon B, Dobrovic A. Lim AM, et al. Clin Epigenetics. 2014 Dec 9;6(1):22. doi: 10.1186/1868-7083-6-22. eCollection 2014. Clin Epigenetics. 2014. PMID: 25859283 Free PMC article.
-
Age-associated hydroxymethylation in human bone-marrow mesenchymal stem cells.
Toraño EG, Bayón GF, Del Real Á, Sierra MI, García MG, Carella A, Belmonte T, Urdinguio RG, Cubillo I, García-Castro J, Delgado-Calle J, Pérez-Campo FM, Riancho JA, Fraga MF, Fernández AF. Toraño EG, et al. J Transl Med. 2016 Jul 8;14(1):207. doi: 10.1186/s12967-016-0966-x. J Transl Med. 2016. PMID: 27393146 Free PMC article.
-
Gervin K, Page CM, Aass HC, Jansen MA, Fjeldstad HE, Andreassen BK, Duijts L, van Meurs JB, van Zelm MC, Jaddoe VW, Nordeng H, Knudsen GP, Magnus P, Nystad W, Staff AC, Felix JF, Lyle R. Gervin K, et al. Epigenetics. 2016 Sep;11(9):690-698. doi: 10.1080/15592294.2016.1214782. Epigenetics. 2016. PMID: 27494297 Free PMC article.
References
-
- Rakyan VK, Down TA, Thorne NP, Flicek P, Kulesha E, Gräf S, Tomazou EM, Bäckdahl L, Johnson N, Herberth M, Howe KL, Jackson DK, Miretti MM, Fiegler H, Marioni JC, Birney E, Hubbard TJP, Carter NP, Tavaré S, Beck S. An integrated resource for genome-wide identification and analysis of human tissue-specific differentially methylated regions (tDMRs). Genome Res. 2008;18:1518–1529. doi: 10.1101/gr.077479.108. - DOI - PMC - PubMed
-
- Irizarry RA, Ladd-Acosta C, Wen B, Wu Z, Montano C, Onyango P, Cui H, Gabo K, Rongione M, Webster M, Ji H, Potash JB, Sabunciyan S, Feinberg AP. The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nat Genet. 2009;41:178–186. doi: 10.1038/ng.298. - DOI - PMC - PubMed
Publication types
MeSH terms
Substances
LinkOut - more resources
Full Text Sources
Other Literature Sources