Biologically-aware Latent Dirichlet Allocation (BaLDA) for the Classification of Expression Microarray

Abstract

Topic models have recently shown to be really useful tools for the analysis of microarray experiments. In particular they have been successfully applied to gene clustering and, very recently, also to samples classification. In this latter case, nevertheless, the basic assumption of functional independence between genes is limiting, since many other a priori information about genes' interactions may be available (co-regulation, spatial proximity or other a priori knowledge). In this paper a novel topic model is proposed, which enriches and extends the Latent Dirichlet Allocation (LDA) model by integrating such dependencies, encoded in a categorization of genes. The proposed topic model is used to derive a highly informative and discriminant representation for microarray experiments. Its usefulness, in comparison with standard topic models, has been demonstrated in two different classification tests.

Biclustering of Expression Microarray Data Using Affinity Propagation

“…One clear example is represented by the topic models [14,5], initially designed for text mining and computer vision applications, and recently successfully applied in the microarray context [23,3,2]. Clearly, the peculiar context may lead to substantial changes in the model so to improve results [21].…”

A Comparison on Score Spaces for Expression Microarray Data Classification

“…PLSA may be very useful in the expression microarray context, since it may provide powerful and interpretable descriptions of experiments [3,22,24]. In particular there is an analogy between the pairs word-document and gene-sample: actually it is reasonable to intend the samples as documents and the genes as words.…”

“…[17,29]. Among others, in recent years some promising techniques were based on a particular class of probabilistic approaches, called topic models, showing optimal and highly interpretable results [2,22,24]. Such probabilistic topic models, the two most famous examples being the Probabilistic Latent Semantic Analysis (PLSA [15]) and the Latent Dirichlet Allocation (LDA [5]), have been imported from the text analysis realm as workhorses in several scientific fields [6,8,33].…”

“…Their wide usage is motivated by their simplicity and expressiveness in dealing with very large datasets both in samples and features number. Therefore, they appeared to be a convenient tool for the microarray data analysis problem, and especially in the context of expression microarray classification [2,22]. Nevertheless, not all the potentialities of these schemes have been exploited in such context.…”

Transfer learning using a nonparametric sparse topic model

“…In particular, we consider learning models for count data, a prominent type of data that arises in bag-of-words representations of text documents, in bioinformatics for example as counts of active genes over pathways, and in other domains. Latent structure in count data has often been modeled with topic models [1], in domains from document collections [2] to bioinformatics [3,4].…”