Classification of heterogeneous microarray data by maximum entropy kernel

[Construction of a High-precision Chemical Prediction System Using Human ESCs]

　Toxicity prediction based on stem cells and tissue derived from stem cells plays a very important role in the fields of biomedicine and pharmacology. Here we report on qRT-PCR data obtained by exposing 20 compounds to human embryonic stem (ES) cells. The data are intended to improve toxicity prediction, per category, of various compounds through the use of support vector machines, and by applying gene networks. The accuracy of our system was 97.5-100% in three toxicity categories: neurotoxins (NTs), genotoxic carcinogens (GCs), and non-genotoxic carcinogens (NGCs). We predicted that two uncategorized compounds (bisphenol-A and permethrin) should be classified as follows: bisphenol-A as a non-genotoxic carcinogen, and permethrin as a neurotoxin. These predictions are supported by recent reports, and as such constitute a good outcome. Our results include two important features: 1) The accuracy of prediction was higher when machine learning was carried out using gene networks and activity, rather than the normal quantitative structure-activity relationship (QSAR); and 2) By using undifferentiated ES cells, the late effect of chemical substances was predicted. From these results, we succeeded in constructing a highly effective and highly accurate system to predict the toxicity of compounds using stem cells.

Prediction of developmental chemical toxicity based on gene networks of human embryonic stem cells

Predictive toxicology using stem cells or their derived tissues has gained increasing importance in biomedical and pharmaceutical research. Here, we show that toxicity category prediction by support vector machines (SVMs), which uses qRT-PCR data from 20 categorized chemicals based on a human embryonic stem cell (hESC) system, is improved by the adoption of gene networks, in which network edge weights are added as feature vectors when noisy qRT-PCR data fail to make accurate predictions. The accuracies of our system were 97.5–100% for three toxicity categories: neurotoxins (NTs), genotoxic carcinogens (GCs) and non-genotoxic carcinogens (NGCs). For two uncategorized chemicals, bisphenol-A and permethrin, our system yielded reasonable results: bisphenol-A was categorized as an NGC, and permethrin was categorized as an NT; both predictions were supported by recently published papers. Our study has two important features: (i) as the first study to employ gene networks without using conventional quantitative structure-activity relationships (QSARs) as input data for SVMs to analyze toxicogenomics data in an hESC validation system, it uses additional information of gene-to-gene interactions to significantly increase prediction accuracies for noisy gene expression data; and (ii) using only undifferentiated hESCs, our study has considerable potential to predict late-onset chemical toxicities, including abnormalities that occur during embryonic development.

An ensemble correlation-based gene selection algorithm for cancer classification with gene expression data

MOTIVATION

Gene selection for cancer classification is one of the most important topics in the biomedical field. However, microarray data pose a severe challenge for computational techniques. We need dimension reduction techniques that identify a small set of genes to achieve better learning performance. From the perspective of machine learning, the selection of genes can be considered to be a feature selection problem that aims to find a small subset of features that has the most discriminative information for the target.

RESULTS

In this article, we proposed an Ensemble Correlation-Based Gene Selection algorithm based on symmetrical uncertainty and Support Vector Machine. In our method, symmetrical uncertainty was used to analyze the relevance of the genes, the different starting points of the relevant subset were used to generate the gene subsets and the Support Vector Machine was used as an evaluation criterion of the wrapper. The efficiency and effectiveness of our method were demonstrated through comparisons with other feature selection techniques, and the results show that our method outperformed other methods published in the literature.