【論】Yeung,2000,Validating cluster for gene express～

Ka Yee Yeung, David R.Haynor, Walter L.Ruzzo.
Validating clustering for gene expression data.
Bioinformatics, 2001 v 17 #4: 309-318.
[PDF][Web Site]

・クラスタリング法のマイクロアレイデータ解析における性能を比較・評価する。
・データ
1.Rat CNS[Wen]
2.Yeast cell cycle[Cho]
3.Human hematopoietic differentiation[Tamayo]
・クラスタリング法
1.CAST
2.Iterative Partition Algorithm
3.K-means
4.Random Clustering ： 性能比較のため
・評価法：FOM

・方法「Our idea is to apply a clustering algorithm to the data from (m-1) conditions, and to use the remaining condition to assess the predictive power of the clustering algorithm.」
・FOMとは「a possible definition of FOM is the average squared distance from the mean expression level in each cluster,」「The aggregate figure of merit, FOM, is an estimate of the total predictive power of the algorithm over all the conditions for k clusters in a data set.」
・結果「None of the three clustering algorithms is a clear winner in our results.」
・目的「Our main contribution is not the comparison of these specific algorithms and metrics, but rather the development of a simple, quantitative data-driven methodology allowing such comparisons to be made between any clustering algorithms and any similarity metrics.」
・わかったこと「we conclude that clustering results with low FOM's tend to have high correspondence to the functional categorization in Wen et al. on the rat CNS data set.」
・結果「No clustering algorithm emerged as a clear-cut winner in this work, and we suggest that flexibility, speed, reliability and ease of implementation may be equally important in differentiating clustering algorithms.」

・やはり(？)、方法比較の場合、なかなかすっきりした結果は出ない。どうしようもないことなのか？
・FOM → 要チェック

《チェック論文》
・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.

【論】Brown,1999,Exploring the new world of the ～

Patrick O.Brown & David Botstein
Exploring the new world of the genome with DNA microarrays
Nature Genetics 21,33-37(1999)
[PDF][Web Site]

・新技術"DNAマイクロアレイ"の紹介。cDNAアレイの仕組み、クラスタリング解析、結果の表示、解釈について。

《チェック論文》
・Cho, R.J. et al. Parallel analysis of genetic selections using whole genome oligonucleotide arrays. Proc. Natl Acad. Sci. USA 95, 3752?3757 (1998).

【論】Kerr,2001,Experimental design for gene ～

M.Kathleen Kerr, Gary A. Churchill,
Experimental design for gene expression microarrays,
Biostatistics, 2:183-201.(2001)
[PDF][Web Site]

・マイクロアレイの実験デザインについて。研究目的に適した結果を得るために、手持ちのアレイデータ同士をどのような組み合わせで比較するとよいか、という問題について。"Block design"のマイクロアレイへの応用。

・概要「We explore the connection between microarray designs and classical block design and use a family of ANOVA models as a guide to choosing a design. We combine principles of good design and A-optimality to give a general set of recommendations for design with microarrays.」

・この"Block design"分野の話が全然わからないので、内容つかめず。

《チェック論文》
・Patrick O.Brown & David Botstein, Exploring the new world of the genome with DNA microarrays, Nature Genetics 21,33-37(1999)

【論】Chen,1997,Ratio-based decisions and the ～

Y. Chen, E. Dougherty, M. Bittner,
Ratio-Based Decisions and the Quantitative Analysis of cDNA Micro-array Images,
Journal of Biomedical Optics, 2:364-374, 1997.
[PDF][Web Site]

・マイクロアレイ関連ではかなり初期の論文。cDNAマイクロアレイの発現量（発現比）の算出法について。画像の色情報をいかに数値に変換するかについて細かく記述している。

・概要「we derive the probability density of the ratio, find the maximum-likelihood estimator for the distribution, and develop an iterative procedure for signal calibration.」

【論】Tavazoie,1999,Systematic determination of ～

Tavazoie S, Hughes JD, Campbell MJ, Cho RJ, Church GM.
Systematic determination of genetic network architecture.
Nat Genet. 1999 Jul;22(3):281-5.
[PDF][Web Site]

・Yeast遺伝子のクラスタリングによる遺伝子機能・ネットワークの推定。
・データ：Yeast、6220遺伝子、時間点15点（Cell cycle 2周期分）、Affy.[Cho]
・クラスタリング法：k-means algorithm（SYSTAT 7.0(SPSS)）、k=30
・Motif検索（AligenACE）

・結果「Most clusters have non-periodic temporal profiles, with some showing complex behaviour.」
・まとめ「The combination of experimental and computational approaches presented here, together with experimental verification of novel motifs and the discovery of their trans-acting factors, should allow the construction of the circuit diagram for the genetic network, allowing us to both understand and manipulate complex cellular processes on a system level.」

・クラスタリング結果をMotifの点から見ているところが目新しい。その後すたれてしまったのだろうか。

【論】Sahara,2002,Comprehensive expression analysis ～

Takehiko Sahara, Takako Goda, and Satoru Ohgiya
Comprehensive Expression Analysis of Time-dependent Genetic Responses in Yeast Cells to Low Temperature
Biol. Chem., Vol.277, Issue 51, 50015-50021, December 20, 2002
[PDF][Web Site]

・Yeastの遺伝子発現量の低温刺激応答について。
・データ：Yeast、5803遺伝子、時間点は低温刺激（30→10℃）後の 0,0.25,0.5,2,4,8[h]、cDNA
・クラスタリング法：Pearson correlation around zero (既存ソフト)
・遺伝子抽出の閾値は発現量変化"2倍"
・解析対象遺伝子は全体の場合と、MIPSデータベースのアノテーションに基づいて、各遺伝子機能ごとに抽出。"transcription","protein synthesis","cell rescue, defense, cell death, and aging","energy and metabolism","signal transduction"の5つのカテゴリーについて。

・「In this experiment, the ratio of fluorescent intensities for these dyes was within 2-fold for 98.5% of all cDNA spots on the microarray. From this result, we defined significant gene expression change as ≧2-fold change in signal intensity.」
・わかったこと「We conclude that the primary and important response to low temperature in yeast is to increase ribosomal complex to compensate for reduced translational ability at low temperature.」
・わかったこと「The result suggests that adaptation mechanisms for low temperature in yast cells are composed of three sequential molecular events.」

【論】Yeung,2001,Model-based clustering and data ～

Yeung KY, Fraley C, Murua A, Raftery AE, Ruzzo WL.
Model-based clustering and data transformations for gene expression data.
Bioinformatics. 2001 Oct;17(10):977-87.
[PDF][Web Site]

・最適なクラスタリング結果を得るための手法について。
・データ：生データ×2、人工データ×3
Gene expression data sets : 1.Ovary data [Schummer], 2.Yeast cell cycle data [Cho]
Synthetic data sets : 1.Mixture of normal distributions based only the ovary data, 2.Randomly resampled ovary data, 3.Cyclic data
・クラスタリング法：
1.EI(equal volume spherical model)
2.VI(unequal volume spherical model)
3.VVV(unconstrained model)
4.EEE(elliptical model)
以上がModel-based clustering、以下は比較した従来法
5.Diagonal
6.CAST
・クラスタリング結果の評価法：1.Average adjusted Rand indices, 2.Average BIC scores

・問題点「Most of the proposed clustering algorithms are largely heuristically motivated, and the issues of determining the 'correct' number of clusters and choosing a 'good' clustering algorithm are not yet rigorously solved.」
・概要「Our contributions include demonstrations of the potential usefulness of the model-based approach by testing the Gaussian mixture assumption for different transformations of expression data, applying existing model-based clustering implementations to both real expression data and symthetic data sets, and comparing the performance of the model-based approach to a leading heuristic-based algorithm.」
・「One of the key advantages of the model-based approach is the availability of a variety of models that distinguish between these scenarios (and others).」
・「Another key advantage of model-based clustering is that there is a principled, data-driven way to approach the model selection problem,」
・結果「Our results show that the BIC analysis not only selects the right model, but also determines the correct number of clusters.」
・教訓「In particular, if the goal is to capture the general patterns across experiments without considering the absolute expression levels, data transformations such as standardization are helpful.」

・論文の構成がハッキリしていて読みやすい。とはいっても理論的な部分はいまいち理解してないのですが。

《チェック論文》
[1]Yeung, Haynor, Ruzzo: Validating Clustering for Gene Expression Data.,Bioinformatics, 2001 v17#4: 309-318. Nat Genet. 1999 Jul;22(3):281-5.
[2]Tavazoie S, Hughes JD, Campbell MJ, Cho RJ, Church GM.,Systematic determination of genetic network architecture.,Nat Genet. 1999 Jul;22(3):213-5.
[3]Chris Fraley and Adrian E. Raftery.,How many clusters? Which clustering method? - Answers via Model-Based Cluster Analysis,Technical, February 1998. Computer Journal 41:578-588 (1998).

【論】Storey,2005,Significance analysis of time ～

Storey JD, Xiao W, Leek JT, Tompkins RG, and Davis RW.
Significance analysis of time course microarray experiments.
Proceedings of the National Academy of Sciences, 102: 12837-12842., 2005
[PDF][Web Site]

・発現量の時間変化に着目した論文
・データ：ヒト（cortex）、35068プローブ×72名（27-92歳）、時間点 0,2,4,6,9,24[hour]、Affy.
・提案手法と、t-test、SAMの結果を比較

・目的「The goal here was to identify genes whose expression changes significantly with respect to age in human kidney cortex tissue.」
・開発ソフト「We have developed a freely available point-and-click software package called EDGE that includes this methodology and new methodology for static experiments.」

・なぜだか非常～～に読みにくい！　肝心の提案している統計解析手法がさっぱりわからない。

【論】Guyon,2002,Gene selection for cancer class～

I.Guyon, J.Weston, S.Barnhill, and V.Vapnik.
Gene selection for cancer classification using support vector machines.
Machine Learning, Vol.46, pp.389--422, 2002
[PDF][Web Site]

・SVMのマイクロアレイデータ解析への応用の代表的な論文。
・データ
その1 Differentiation of two types of Leukemia, 72サンプル×7129遺伝子[Golub]
その2 Colon cancer diagnosis, 62サンプル×2000遺伝子[Alon]
・クラス分け法
1.SVM RFE
2.Linear Discriminant Analysis RFE
3.Mean Squared Error (Pseudo-inverse) RFE
4.Baseline method
・クラス分け評価法：Leave-one-out success rate

・概要「In this paper we investigate pruning techniques that eliminate some of the original input features and retain a minimum subset of features that yield best classification performance.」
・従来法「Each coefficient w_i is computed with information about a single feature (gene) and does not take into account mutual information between features.」
・結果「All our feature selection experiments using various classifiers (SVM, LDA, MSE) indicated that better features are obtained by using RFE than by using the weights of a single classifier. Similarly, better results are obtained by eliminating one feature at a time than by eliminating chunks of features. However, there are only significant differences for the smaller subset of genes (less than 100).」
・展望「We envision that linear classifiers are going to continue to play an important role in the analysis of DNA micro-array because of the large ratio number of features over number of training patterns.」
・Golubが使用した相関係数　w_i = (μ_i(+) - μ_i(-))/(σ_i(+) + σ_i(-))
・Fisher's discriminant criterion (μ_i(+) - μ_i(-))²/(σ_i(+)² + σ_i(-)²)

【論】Cho,1998,A genome-wide transcriptional ～

Cho RJ, Campbell MJ, Winzeler EA, Steinmetz L, Conway A, Wodicka L, Wolfsberg TG, Gabrielian AE, Landsman D, Lockhart DJ, Davis RW.
A genome-wide transcriptional analysis of the mitotic cell cycle.
Mol Cell. 1998 Jul;2(1):65-73.
[PDF]

・酵母の細胞周期に着目したアレイ解析。マイクロアレイを使用したものとしては初期の論文。
・細胞周期の4期（G1, S, G2, M）にそれぞれ特異的に発現する遺伝子を抽出。

・展望「Because more than 20% of human proteins display significant homology to yeast proteins, these results aslo link a range of human genes to cell cycle period-specific biological functions.」