【論】Schadt,2001,Feature extraction and normalizati～

Eric E.Schadt, Cheng Li, Byron Ellis, and Wing H.Wong
Feature extraction and normalization algorithms for high-density oligonucleotide gene expression array data
Journal of Cellular Biochemistry Supplement 37:120-125(2001)
[PDF][Web Site]

・マイクロアレイデータ解析用ソフト、DNA-Chip Analyzer(dChip)の性能を検証する。
・発現量数値化法（Feature extraction）の比較：APS (adaptive pixel selection) algorithm と TP75 (75th percentile) algorithm (従来法)を比較。
・正規化法の比較
1.Linear normalization (LN)
2.Smoothing spline normalization (GCVSS)
3.Invariant difference selection/smoothing spline (IDS) methods （←提案法）

【論】Li,2001,Model-based analysis of oligonucleotid～

Cheng Li and Wing Hung Wong
Model-based analysis of oligonucleotide arrays: model validation, design issues and standard error application
Genome Biology 2(8): research0032.1-0032.11(2001)
[PDF][Web Site]

・オリゴヌクレオチドアレイ（Affymetrix社 GeneChip）の信頼度を評価した基礎的な研究。（？）
・使用ソフト：dChip

・概要「Here we investigate the stability of the probe-sensitivity index across different tissue types, the reproducibility of results in replicate experiments, and the use of MBEI in perfect match (PM)-only arrays.」

・さっぱり内容がつかめない。辞書的な意味は分かっても、文脈的によく分からない言葉が多い→「model-based」「standard error」「PM-only arrays」

【論】Irizarry,2003,Summaries of Affymetrix GeneChip～

Rafael A. Irizarry, Benjamin M. Bolstad, Francois Collin, Leslie M. Cope, Bridget Hobbs and Terence P. Speed
Summaries of Affymetrix GeneChip probe level data
Nucleic Acids Research. Vol.31, No.4 e15(2003)
[PDF][Web Site]

・発現量を示す画像データから数値データを算出する手法について。MAS5.0, dChip, RMAの性能比較。
・評価指標
(i) the precision of the measures of expression, as estimated by standard deviations across replicate chips
(ii) the consistency of fold change estimates based on widely differing concentrations of target mRNA hybridized to the chip
(iii) the specificity and sensitivity of the measures' ability to detect differential expression, presented in terms of receiver operating charasteristic (ROC) curves

・問題点「To define a measure of expression representing the amount of the corresponding mRNA species it is necessary to summarize probe intensities for each probe set.」
・問題点「There is no gold standard to compare and test summaries of probe level data.」
・結果「This assessment demonstrated that RMA provides more consistent estimates of fold change.」
・結果「In this comparison, dChip performed almost as well as RMA and singnificantly better than MAS5.0. This assessment demonstrated that using RMA provides higher specificity and sensitivity when using fold change analysis to detect differential expression.」
・まとめ「Specifically we found that: (i) RMA has better precision; in particular, for lower expression values we found that RMA provides a greater than 5-fold reduction of the within-replicate variance as compared to dChip and MAS5.0; (ii) RMA provided more consistent estimates of fold change; (iii) RMA provided higher specificity when using fold change analysis to detect differential expression.」

【論】Workman,2002,A new non-linear normalization ～

Christopher Workman, Lars Juhl Jensen, Hanne Jarmer, Randy Berka, Laurent Gautier, Henrik Bjørn Nielser, Hans-Henrik Saxild, Claus Nielsen, Søren Brunak and Steen Knudsen
A new non-linear normalization method for reducing variability in DNA microarray experiments
Genome Biology 2002, 3:research0048.1-0048.16
[PDF][Web Site]

・マイクロアレイデータの新しい正規化の方法を提案する。
・実験対象のマイクロアレイをcDNAかGeneChipのどちらかに限定する場合が多いが、この論文では『両方に使えます』というのがウリ。
・データ
1.HIV-infected T-cells, 6サンプル, GeneChip
2.B.subtilis, 6サンプル, cDNA
・正規化法
1.Lowess(robust local-linear regression)
2.Invariant set
3.Qspline：提案法, R+Bioconductorに実装
4.Spatial
ほか

・結果「The applications of these methods to oligonucleotide arrays reduced the relative error between replicates by 5-10% compared with a standard global normalization method. Application to cDNA arrays showed improvements over the standard method and over Cy3-Cy5 normalization based on dye-swap replication.」
・問題点「Much of the literature addressing microarray normalization concerns cDNA array data, whereas only a few examples can be found for oligonucleotide arrays.」
・概要「Here we review microarray normalization techniques, present the qspline method and show its application to Affymetrix oligonucleotide arrays of human T-cell cultures and cDNA arrays of Bacillus subtilis and Arabidopsis thaliana.」
・目的「The goal of signal-dependent normalization is to make signal distributions comparable across the intensity range.」

【論】Reiner,2003,Identifying differentially express～

Anat Reiner, Daniel Yekutieli and Yoav Benjamini
Identifying differentially expressed genes using false discovery rate contolling procedures
Bioinformatics, 19(3), 368-375, 2003.
[PDF][Web Site]

・マイクロアレイデータを用いたFDRの性能評価。
・データ：マウス[Dudoit,2002]
・解析法
1.Resampling upper limit (FDR)
2.Resampling point-estimate (FDR)
3.BH point-estimate (FDR)
4.BH LSU - No Resampling (FDR)
5.Westfall-Young (FWE)

・FDRとは「The FDR is the expected proportion of erroneously rejected null hypotheses among the rejected ones.」
・目的「The goal of the experiment was to identify genes with altered expression in the livers of mice with very low HDL cholesterol levels compared to inbred control mice.」

・そう難しいことをやっているわけではなさそうだけど、何をどうしているのか、さっぱり内容つかめず。お手上げ。「FDRをcontrolする」イメージが湧かない。

【論】Hoffmann,2002,Profound effect of normalization～

Reinhard Hoffmann, Thomas Seidl and Martin Dugas
Profound effect of normalization on detection of differentially expressed genes in oligonucleotide microarray data analysis
Genome Biol 2002, 3:RESEARCH0033
[PDF][Web Site]

・マイクロアレイデータの正規化法による解析結果の違いを比較する。
・データ：マウス, B-cell precursor gene-expression dataset[Hoffmann]
・正規化法
1.Global scaling
2.Invariant set
3.Invariant feature(MBEV, model-based expression values)
4.Invariant feature(AD, average difference)
・統計処理法
1.F, F-test for parametric ANOVA
2.KW, H(Kruskal-Wallis) test for nonparametric ANOVA
3.SAM, significance analysis of microarrays
・有意な発現差を示す遺伝子の数（割合）で評価する。

・概要「We have employed four different normalization methods and all possible combinations with three different statistical algorithms for detection of differentially expressed genes on a prototype dataset.」
・結論「Normalization has a profound influence of detection of differentially expressed genes. This influence is higher than that of three subsequent statistical analysis procedures examined.」
・問題点「The question naturally arises of which combination of algorithms is 'best' for analyzing gene-expression data. There is probably no general answer.」

チェック論文
・A. D. Long, H. J. Mangalam, B. Y. P. Chan, L. Tolleri, G. W. Hatfield, and P. Baldi, Improved Statistical Inference from DNA Microarray Data Using Analysis of Variance and A Bayesian Statistical Framework., J. Biol. Chem., June 1, 2001; 276(23): 19937-19944.

【論】Tamayo,1999,Interpreting patterns of gene expr～

Tamayo P, Slonim D, Mesirov J, Zhu Q, Kitareewan S, Dmitrovsky E, Lander ES, Golub TR.
Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation.
Proc Natl Acad Sci U S A. 1999 Mar 16;96(6):2907-12.
[PDF]

・SOM（自己組織化マップ）のマイクロアレイデータ解析への応用。
・SOM実行ソフト：GENECLUSTER
・データ1：Yeast、スタンフォード大のWebSiteよりダウンロード
・データ2：Human、Hematopoietic defferentiation（血液生成分化？？）、約6000遺伝子、HL-60・U937・Jurkat・NB4の四つのモデル

・データ前処理「Genes were eliminated if they did not show a relative change of x and an absolute change of y units, with (x,y)=(2,35) for yeast data and (x,y)=(3,100) for human data. Expression levels were then normalized to have mean=0 and variance=1.」

・なぜだか内容が非常～～につかみにくい。

【論】Cope,2004,A benchmark of Affymetricx GeneChip～

Leslie M. Cope, Rafael A. Irizarry, Harris A. Jaffee, Zhijin Wu and Terence P. Speed
A benchmark for Affymetrix GeneChip expression measures
Bioinformatics. 2004 Feb 12;20(3):323-31.
[PDF]

・Affymetrix社製GeneChipのデータ生成法を比較し、各方法の特性を明らかにする。結果はDilution studyとSpike-in studyのBenchmark dataとして応用する。
・データ
1.The dilution study by GeneLogic
2.The spike-in study by Affymetrix
・解析ソフト
1.Affymetrix software
2.R + Bioconductor
・データ生成法
1.MAS5.0
2.dChip
3.RMA(robust multi-array analysis)
4.RMA(Not multi-array)
5.RMA(Not robust)
6.RMA(MM as PM)
・評価法（グラフ）
1.MA plot
2.Variance across replicates plot
3.Sensitivity of expression ratios to total quantity of RNA plot
4.Observed expression versus nominal expression plots
5.ROC curves
6.Observed fold-change versus nominal fold-change plots
7.Predicted variability plot
・実験
その1.データ生成法1～4を比較
その2.データ生成法4～6を比較

・ROCとは「Receiver Operator Charasteristic (ROC) curves offer a graphical representation of both specificity and sensitivity for such a rule. ROC curves are created by plotting the true positive (TP) rate (sensitivity) against false positive (FP) rate (1-specificity) obtained at each possible threshold value.」
・目的「This is because our aim is the comparison of bias and variance of measurement processes, not algorithms for the specific task of identifying defferential expression.」

・キーワードである「Dilution study」と「Spike-in study」が、何のことやらわからん。

《チェック論文》
・Irizarry,R.A., Bolstad, B.M., Collin,F., Cope,L.M., Hobbs,B. and Speed,T.P. (2003b) Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res., 31, e15.
・Workman,C., Jensen,L.J., Jarmer,H., Berka,R., Gautier,L., Nielsen,H.B., Saxild,H., Nielsen,C., Brunak,S. and Knudsen,S. (2002) A new non-linear normalization method for reducing variability in DNA microarray experiments. Genome Biol., 3, research 0048.

【論】Cho,1998,Parallel analysis of genetic selectio～

Raymond J. Cho, Micheline Fromont-Racine, Lisa Wodicka, Becket Feierbach, Timothy Stearns, Pierre Legrain, David J. Lockhart, and Ronald W. Davis
Parallel analysis of genetic selections using whole genome oligonucleotide arrays.
Proc. Natl Acad. Sci. USA 95, 3752-3757 (1998).
[PDF][Web Site]

・DNAマイクロアレイ分野の初期における、Affymetrixのチップを使った基礎的な研究。チップを作成し、遺伝子発現量をうまく測定できるかを検証する。

・目的「We present an alternative approach in which identification of nucleic acids is accomplished by direct hybridization to high-density oligonucleotide arrays.」
・利点「The use of array hybridization to analyze results from these screens would eliminate the need to maintain large numbers of individual clones in tissue culture until they can be sequenced.」

【論】Shedden,2005,Comparison of seven methods for ～

Kerby Shedden, Wei Chen, Rork Kuick, Debashis Ghosh, James Macdonald, Kathleen R Cho, Thomas J Giordano, Stephen B Gruber, Eric R Fearon, Jeremy MG Taylor, and Samir Hanash
Comparison of seven methods for producing Affymetrix expression scores based on False Discovery Rates in disease profiling data
BMC Bioinformatics. 2005; 6: 26. Published online 2005 February 10. doi: 10.1186/1471-2105-6-26.
[PDF][Web Site]

・Affymetrix社製GeneChipについて、画像データを数値データに変換する手法の比較をする。
・データ
1.Ovary tumor ： 79サンプル
2.Colon tumor ： 47サンプル
・処理法
1.Dchip ： この分野の初期の頃から使用されてきた
2.GCRMA-EB
3.GCRMA-MLE
4.MAS5 ： Affymetrix標準の処理法
5.PDNN
6.RMA
7.TM(trimmed mean) ： まだあまり知られていない方法
・評価法 ： 上記7つの方法について、それぞれt-test statisticとWilcoxon rank-sum statistic（Mann-Whitney statistic）をそれぞれ計算しFDRで評価する

・目的「Our goal here was to carry out a comparative study of Affymetrix array processing methods using data sets from typical biological experiments seeking differentially expressed genes in human tissue samples.」
・結果「Based on an overall measure of performance, two of the seven methods (Dchip and a trimmed mean approach) are superior in the two data sets considered here. Two other methods (MAS5 and GCRMA-EB) are inferior, while results for the other methods are mixed.」

《チェック論文》
・Hoffmann R, Seidl T, Dugas M: Profound effect of normalization on detection of differentially expressed genes in oligonucleotide microarray data analysis. Genome Biol 2002, 3:RESEARCH0033.