Dataset Information

Biomarker signature identification in "omics" data with multi-class outcome.

ABSTRACT: Biomarker signature identification in "omics" data is a complex challenge that requires specialized feature selection algorithms. The objective of these algorithms is to select the smallest set(s) of molecular quantities that are able to predict a given outcome (target) with maximal predictive performance. This task is even more challenging when the outcome comprises of multiple classes; for example, one may be interested in identifying the genes whose expressions allow discrimination among different types of cancer (nominal outcome) or among different stages of the same cancer, e.g. Stage 1, 2, 3 and 4 of Lung Adenocarcinoma (ordinal outcome). In this work, we consider a particular type of successful feature selection methods, named constraint-based, local causal discovery algorithms. These algorithms depend on performing a series of conditional independence tests. We extend these algorithms for the analysis of problems with continuous predictors and multi-class outcomes, by developing and equipping them with an appropriate conditional independence test procedure for both nominal and ordinal multi-class targets. The test is based on multinomial logistic regression and employs the log-likelihood ratio test for model selection. We present a comparative, experimental evaluation on seven real-world, high-dimensional, gene-expression datasets. Within the scope of our analysis the results indicate that the new conditional independence test allows the identification of smaller and better performing signatures for multi-class outcome datasets, with respect to the current alternatives for performing the independence tests.

SUBMITTER: Lagani V

PROVIDER: S-EPMC3962136 | biostudies-literature | 2013

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Biomarker signature identification in "omics" data with multi-class outcome.

Lagani Vincenzo V Kortas George G Tsamardinos Ioannis I

Computational and structural biotechnology journal 20130608

Biomarker signature identification in "omics" data is a complex challenge that requires specialized feature selection algorithms. The objective of these algorithms is to select the smallest set(s) of molecular quantities that are able to predict a given outcome (target) with maximal predictive performance. This task is even more challenging when the outcome comprises of multiple classes; for example, one may be interested in identifying the genes whose expressions allow discrimination among diff ...[more]

PMID: 24688712

Similar Datasets

Project description:PurposeTo identify a gene signature for the prognosis of breast cancer using high-throughput analysis.MethodsRNASeq, single nucleotide polymorphism (SNP), copy number variation (CNV) data and clinical follow-up information were downloaded from The Cancer Genome Atlas (TCGA), and randomly divided into training set or verification set. Genes related to breast cancer prognosis and differentially expressed genes (DEGs) with CNV or SNP were screened from training set, then integrated together for feature selection of identify robust biomarkers using RandomForest. Finally, a gene-related prognostic model was established and its performance was verified in TCGA test set, Gene Expression Omnibus (GEO) validation set and breast cancer subtypes.ResultsA total of 2287 prognosis-related genes, 131 genes with amplified copy numbers, 724 gens with copy number deletions, and 280 genes with significant mutations screened from Genomic Variants were closely correlated with the development of breast cancer. A total of 120 candidate genes were obtained by integrating genes from Genomic Variants and those related to prognosis, then 6 characteristic genes (CD24, PRRG1, IQSEC3, MRGPRX, RCC2, and CASP8) were top-ranked by RandomForest for feature selection, noticeably, several of these have been previously reported to be associated with the progression of breast cancer. Cox regression analysis was performed to establish a 6-gene signature, which can stratify the risk of samples from training set, test set and external validation set, moreover, the five-year survival AUC of the model in the training set and validation set was both higher than 0.65. Thus, the 6-gene signature developed in the current study could serve as an independent prognostic factor for breast cancer patients.ConclusionThis study constructed a 6-gene signature as a novel prognostic marker for predicting the survival of breast cancer patients, providing new diagnostic/prognostic biomarkers and therapeutic targets for breast cancer patients.

Project description:BackgroundThe tumor immune microenvironment significantly affects tumor occurrence, progression, and prognosis, but its impact on the prognosis of low-grade glioma (LGG) patients with epilepsy has not been reported. Hence, the purpose of this study is to explore its effect on LGG patients with epilepsy.MethodsThe data of LGG patients derived from the TCGA database. The level of immune cell infiltration and the proportion of 22 immune cells were evaluated by ESTIMATE and CIBERSORT algorithms, respectively. The Cox and LASSO regression analysis was adopted to determine the DEGs, and further established the clustering and risk score models. The association between genomic alterations and risk score was investigated using CNV and somatic mutation data. GSVA was adopted to identify the immunological pathways, immune infiltration and inflammatory profiles related to the signature genes. The Tumor Immune Dysfunction and Exclusion (TIDE) algorithm and GDSC database were used to predict the patient's response to immunotherapy and chemotherapy, respectively.ResultsThe prognosis of LGG patients with epilepsy was associated with the immune score. Three prognostic DEGs (ABCC3, PDPN, and INA) were screened out. The expression of signature genes was regulated by DNA methylation. The clustering and risk score models could stratify glioma patients into distinct prognosis groups. The risk score was an independent predictor in prognosis, with a high risk-score indicating a poor prognosis, more malignant clinicopathological and genomic aberration features. The nomogram had the better predictive ability. Patients at high risk had a higher level of macrophage infiltration and increased inflammatory activities associated with T cells and macrophages. While the higher percentage of NK CD56bright cell and more active inflammatory activity associated with B cell were present in the low-risk patients. The signature genes participated in the regulation of immune-related pathways, such as IL6-JAK-STAT3 signaling, IFN-α response, IFN-γ response, and TNFA-signaling-via-NFKB pathways. The high-risk patients were more likely to benefit from anti-PD1 and temozolomide (TMZ) treatment.ConclusionAn immune-related gene signature was established based on ABCC3, PDPN, and INA, which can be used to predict the prognosis, immune infiltration status, immunotherapy and chemotherapy response of LGG patients with epilepsy.

Dataset Information

Biomarker signature identification in "omics" data with multi-class outcome.

Publications

Biomarker signature identification in "omics" data with multi-class outcome.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets