Judging the quality of gene expression-based clustering methods using gene annotation.

By Francis D. Gibbons, Frederick P. Roth. Genome Research, 2002

378 citations recorded by the source corpus.

We compare several commonly used expression-based gene clustering algorithms using a figure of merit based on the mutual information between cluster membership and known gene attributes. By studying various publicly available expression data sets we conclude that enrichment of clusters for biological function is, in general, highest at rather low cluster numbers. As a measure of dissimilarity between the expression patterns of two genes, no method outperforms Euclidean distance for ratio-based measurements, or Pearson distance for non-ratio-based measurements at the optimal choice of cluster number. We show the self-organized-map approach to be best for both measurement types at higher numbers of clusters. Clusters of genes derived from single- and average-linkage hierarchical clustering tend to produce worse-than-random results.

Subfields: Molecular Biology

View Paper: https://www.semanticscholar.org/paper/190666fde9b83171f7bf1a560a3ed7d31596355e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

Citations in the corpus

LSimpute: accurate estimation of missing values in microarray data with least squares methods. (2004, methodology): “…imputation methods on data sets that do include missing values, clustering can be done based on data obtained using the alternative imputation methods and the results evaluated for instance using the method suggested by Gibbons and Roth (16) for comparing gene annotations with clustering systems.”
Information criterion-based clustering with order-restricted candidate profiles in short time-course microarray experiments (2009, positive): “Among the recently developed methods for assessing clustering reliability [34-38], we adopt a general bootstrap framework proposed by Kerr and Churchill [36], in which the clustering procedure is first applied to the original data and then to a large number of bootstrap samples obtained from perturb”
Data-analysis strategies for image-based cell profiling (2017, methodology): “Other supervised prediction models can also be used to learn relations between morphological features and biological activity assays, such as Bayesian matrix factorization, neural networks, and random forests 121 .”
How does gene expression clustering work? (2005, methodology): “For example, if our goal is to cluster together genes with similar function, then we can use existing functional annotations to verify how well that goal has been achieved 8,9 .”
Kernel Methods and Machine Learning (2014, methodology): “In addition, according to Gibbons and Roth [77], the SOM reportedly holds advantages over K-means when the number of clusters is large.”
Gene regulatory network inference: Data integration in dynamic models - A review (2009, background): “Nevertheless, a common subsequent analysis step is to annotate each cluster with a functional category that is representative for that cluster (Gibbons and Roth, 2002).”
Characterizing gene sets with FuncAssociate (2003, background): “While such corrections may be appropriate when the multiple hypotheses are independent, they can be misleading when there is substantial dependence among the hypotheses, as is the case here given the dependence between GO attributes (Gibbons and Roth, 2002).”