Dataset Information

Functional analysis of multiple genomic signatures demonstrates that classification algorithms choose phenotype-related genes.

ABSTRACT: Gene expression signatures of toxicity and clinical response benefit both safety assessment and clinical practice; however, difficulties in connecting signature genes with the predicted end points have limited their application. The Microarray Quality Control Consortium II (MAQCII) project generated 262 signatures for ten clinical and three toxicological end points from six gene expression data sets, an unprecedented collection of diverse signatures that has permitted a wide-ranging analysis on the nature of such predictive models. A comprehensive analysis of the genes of these signatures and their nonredundant unions using ontology enrichment, biological network building and interactome connectivity analyses demonstrated the link between gene signatures and the biological basis of their predictive power. Different signatures for a given end point were more similar at the level of biological properties and transcriptional control than at the gene level. Signatures tended to be enriched in function and pathway in an end point and model-specific manner, and showed a topological bias for incoming interactions. Importantly, the level of biological similarity between different signatures for a given end point correlated positively with the accuracy of the signature predictions. These findings will aid the understanding, and application of predictive genomic signatures, and support their broader application in predictive medicine.

SUBMITTER: Shi W

PROVIDER: S-EPMC2920075 | biostudies-literature | 2010 Aug

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Functional analysis of multiple genomic signatures demonstrates that classification algorithms choose phenotype-related genes.

Shi W W Bessarabova M M Dosymbekov D D Dezso Z Z Nikolskaya T T Dudoladova M M Serebryiskaya T T Bugrim A A Guryanov A A Brennan R J RJ Shah R R Dopazo J J Chen M M Deng Y Y Shi T T Jurman G G Furlanello C C Thomas R S RS Corton J C JC Tong W W Shi L L Nikolsky Y Y

The pharmacogenomics journal 20100801 4

Gene expression signatures of toxicity and clinical response benefit both safety assessment and clinical practice; however, difficulties in connecting signature genes with the predicted end points have limited their application. The Microarray Quality Control Consortium II (MAQCII) project generated 262 signatures for ten clinical and three toxicological end points from six gene expression data sets, an unprecedented collection of diverse signatures that has permitted a wide-ranging analysis on ...[more]

PMID: 20676069

Similar Datasets

Project description:Pearson's correlation coefficient (ρ) is the most commonly reported metric of the success of prediction in genomic selection (GS). However, in real breeding ρ may not be very useful for assessing the quality of the regression in the tails of the distribution, where individuals are chosen for selection. This research used 14 maize and 16 wheat data sets with different trait-environment combinations. Six different models were evaluated by means of a cross-validation scheme (50 random partitions each, with 90% of the individuals in the training set and 10% in the testing set). The predictive accuracy of these algorithms for selecting individuals belonging to the best α=10, 15, 20, 25, 30, 35, 40% of the distribution was estimated using Cohen's kappa coefficient (κ) and an ad hoc measure, which we call relative efficiency (RE), which indicates the expected genetic gain due to selection when individuals are selected based on GS exclusively. We put special emphasis on the analysis for α=15%, because it is a percentile commonly used in plant breeding programmes (for example, at CIMMYT). We also used ρ as a criterion for overall success. The algorithms used were: Bayesian LASSO (BL), Ridge Regression (RR), Reproducing Kernel Hilbert Spaces (RHKS), Random Forest Regression (RFR), and Support Vector Regression (SVR) with linear (lin) and Gaussian kernels (rbf). The performance of regression methods for selecting the best individuals was compared with that of three supervised classification algorithms: Random Forest Classification (RFC) and Support Vector Classification (SVC) with linear (lin) and Gaussian (rbf) kernels. Classification methods were evaluated using the same cross-validation scheme but with the response vector of the original training sets dichotomised using a given threshold. For α=15%, SVC-lin presented the highest κ coefficients in 13 of the 14 maize data sets, with best values ranging from 0.131 to 0.722 (statistically significant in 9 data sets) and the best RE in the same 13 data sets, with values ranging from 0.393 to 0.948 (statistically significant in 12 data sets). RR produced the best mean for both κ and RE in one data set (0.148 and 0.381, respectively). Regarding the wheat data sets, SVC-lin presented the best κ in 12 of the 16 data sets, with outcomes ranging from 0.280 to 0.580 (statistically significant in 4 data sets) and the best RE in 9 data sets ranging from 0.484 to 0.821 (statistically significant in 5 data sets). SVC-rbf (0.235), RR (0.265) and RHKS (0.422) gave the best κ in one data set each, while RHKS and BL tied for the last one (0.234). Finally, BL presented the best RE in two data sets (0.738 and 0.750), RFR (0.636) and SVC-rbf (0.617) in one and RHKS in the remaining three (0.502, 0.458 and 0.586). The difference between the performance of SVC-lin and that of the rest of the models was not so pronounced at higher percentiles of the distribution. The behaviour of regression and classification algorithms varied markedly when selection was done at different thresholds, that is, κ and RE for each algorithm depended strongly on the selection percentile. Based on the results, we propose classification method as a promising alternative for GS in plant breeding.

Project description:BACKGROUND:The experience with running various types of classification on the CAMDA neuroblastoma dataset have led us to the conclusion that the results are not always obvious and may differ depending on type of analysis and selection of genes used for classification. This paper aims in pointing out several factors that may influence the downstream machine learning analysis. In particular those factors are: type of the primary analysis, type of the classifier and increased correlation between the genes sharing a protein domain. They influence the analysis directly, but also interplay between them may be important. We have compiled the gene-domain database and used it for analysis to see the differences between the genes that share a domain versus the rest of the genes in the datasets. RESULTS:The major findings are: pairs of genes that share a domain have an increased Spearman's correlation coefficients of counts; genes sharing a domain are expected to have a lower predictive power due to increased correlation. For most of the cases it can be seen with the higher number of misclassified samples; classifiers performance may vary depending on a method, still in most cases using genes sharing a domain in the training set results in a higher misclassification rate; increased correlation in genes sharing a domain results most often in worse performance of the classifiers regardless of the primary analysis tools used, even if the primary analysis alignment yield varies. CONCLUSIONS:The effect of sharing a domain is likely more a results of real biological co-expression than just sequence similarity and artifacts of mapping and counting. Still, this is more difficult to conclude and needs further research. The effect is interesting itself, but we also point out some practical aspects in which it may influence the RNA sequencing analysis and RNA biomarker use. In particular it means that a gene signature biomarker set build out of RNA-sequencing results should be depleted for genes sharing common domains. It may cause to perform better when applying classification. REVIEWERS:This article was reviewed by Dimitar Vassiliev and Susmita Datta.

Project description:ObjectiveThe purpose of this study was to develop an algorithm incorporating MRI metrics to classify patients with mild traumatic brain injury (mTBI) and controls.MethodsThis was an institutional review board-approved, Health Insurance Portability and Accountability Act-compliant prospective study. We recruited patients with mTBI and healthy controls through the emergency department and general population. We acquired data on a 3.0T Siemens Trio magnet including conventional brain imaging, resting-state fMRI, diffusion-weighted imaging, and magnetic field correlation (MFC), and performed multifeature analysis using the following MRI metrics: mean kurtosis (MK) of thalamus, MFC of thalamus and frontal white matter, thalamocortical resting-state networks, and 5 regional gray matter and white matter volumes including the anterior cingulum and left frontal and temporal poles. Feature selection was performed using minimal-redundancy maximal-relevance. We used classifiers including support vector machine, naive Bayesian, Bayesian network, radial basis network, and multilayer perceptron to test maximal accuracy.ResultsWe studied 24 patients with mTBI and 26 controls. Best single-feature classification uses thalamic MK yielding 74% accuracy. Multifeature analysis yields 80% accuracy using the full feature set, and up to 86% accuracy using minimal-redundancy maximal-relevance feature selection (MK thalamus, right anterior cingulate volume, thalamic thickness, thalamocortical resting-state network, thalamic microscopic MFC, and sex).ConclusionMultifeature analysis using diffusion-weighted imaging, MFC, fMRI, and volumetrics may aid in the classification of patients with mTBI compared with controls based on optimal feature selection and classification methods.Classification of evidenceThis study provides Class III evidence that classification algorithms using multiple MRI features accurately identifies patients with mTBI as defined by American Congress of Rehabilitation Medicine criteria compared with healthy controls.

Dataset Information

Functional analysis of multiple genomic signatures demonstrates that classification algorithms choose phenotype-related genes.

Publications

Functional analysis of multiple genomic signatures demonstrates that classification algorithms choose phenotype-related genes.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets