Dataset Information

Diagnostic biases in translational bioinformatics.

ABSTRACT: With the surge of translational medicine and computational omics research, complex disease diagnosis is more and more relying on massive omics data-driven molecular signature detection. However, how to detect and prevent possible diagnostic biases in translational bioinformatics remains an unsolved problem despite its importance in the coming era of personalized medicine.In this study, we comprehensively investigate the diagnostic bias problem by analyzing benchmark gene array, protein array, RNA-Seq and miRNA-Seq data under the framework of support vector machines for different model selection methods. We further categorize the diagnostic biases into different types by conducting rigorous kernel matrix analysis and provide effective machine learning methods to conquer the diagnostic biases.In this study, we comprehensively investigate the diagnostic bias problem by analyzing benchmark gene array, protein array, RNA-Seq and miRNA-Seq data under the framework of support vector machines. We have found that the diagnostic biases happen for data with different distributions and SVM with different kernels. Moreover, we identify total three types of diagnostic biases: overfitting bias, label skewness bias, and underfitting bias in SVM diagnostics, and present corresponding reasons through rigorous analysis. Compared with the overfitting and underfitting biases, the label skewness bias is more challenging to detect and conquer because it can be easily confused as a normal diagnostic case from its deceptive accuracy. To tackle this problem, we propose a derivative component analysis based support vector machines to conquer the label skewness bias by achieving the rivaling clinical diagnostic results.Our studies demonstrate that the diagnostic biases are mainly caused by the three major factors, i.e. kernel selection, signal amplification mechanism in high-throughput profiling, and training data label distribution. Moreover, the proposed DCA-SVM diagnosis provides a generic solution for the label skewness bias overcome due to the powerful feature extraction capability from derivative component analysis. Our work identifies and solves an important but less addressed problem in translational research. It also has a positive impact on machine learning for adding new results to kernel-based learning for omics data.

SUBMITTER: Han H

PROVIDER: S-EPMC4522082 | biostudies-literature | 2015

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Diagnostic biases in translational bioinformatics.

Han Henry H

BMC medical genomics 20150801

<h4>Background</h4>With the surge of translational medicine and computational omics research, complex disease diagnosis is more and more relying on massive omics data-driven molecular signature detection. However, how to detect and prevent possible diagnostic biases in translational bioinformatics remains an unsolved problem despite its importance in the coming era of personalized medicine.<h4>Methods</h4>In this study, we comprehensively investigate the diagnostic bias problem by analyzing benc ...[more]

PMID: 26232237

Similar Datasets

Project description:The subjective measures used to study mood disorders in humans cannot be replicated in animals; however, the increasing application of objective neuropsychological methods provides opportunities to develop translational animal tasks. Here we describe a novel behavioral approach, which has enabled us to investigate similar affective biases in rodents. In our affective bias test (ABT), rats encounter two independent positive experiences--the association between food reward and specific digging substrate--during discrimination learning sessions. These are performed on separate days under either neutral conditions or during a pharmacological or affective state manipulation. Affective bias is then quantified using a preference test where both previously rewarded substrates are presented together and the rat's choices recorded. The absolute value of the experience is kept consistent and all other factors are counterbalanced so that any bias at recall can be attributed to treatment. Replicating previous findings from studies in healthy volunteers, we observe significant positive affective biases following acute treatment with typical (fluoxetine, citalopram, reboxetine, venlafaxine, clomipramine) and atypical antidepressants (agomelatine, mirtazapine), and significant negative affective biases following treatment with drugs associated with inducing negative affective states in humans (FG7142, rimonabant, 13-cis retinoic acid). We also observed that acute psychosocial stress and environmental enrichment induce significant negative and positive affective biases, respectively, and provide evidence that these affective biases involve memory consolidation. The positive and negative affective biases induced in our test also mirror the antidepressant and pro-depressant effects of these drugs in patients suggesting our test has both translational and predictive validity. Our results suggest that cognitive affective biases could contribute to drug- or stress-induced mood changes in people and support the hypothesis that a cognitive neuropsychological mechanism contributes to antidepressant drug efficacy.

Project description:Extracting biologically meaningful information from the continuing flood of genomic data is a major challenge in the life sciences. Codon usage bias (CUB) is a general feature of most genomes and is thought to reflect the effects of both natural selection for efficient translation and mutation bias. Here we present a mechanistically interpretable, Bayesian model (ribosome overhead costs Stochastic Evolutionary Model of Protein Production Rate [ROC SEMPPR]) to extract meaningful information from patterns of CUB within a genome. ROC SEMPPR is grounded in population genetics and allows us to separate the contributions of mutational biases and natural selection against translational inefficiency on a gene-by-gene and codon-by-codon basis. Until now, the primary disadvantage of similar approaches was the need for genome scale measurements of gene expression. Here, we demonstrate that it is possible to both extract accurate estimates of codon-specific mutation biases and translational efficiencies while simultaneously generating accurate estimates of gene expression, rather than requiring such information. We demonstrate the utility of ROC SEMPPR using the Saccharomyces cerevisiae S288c genome. When we compare our model fits with previous approaches we observe an exceptionally high agreement between estimates of both codon-specific parameters and gene expression levels ([Formula: see text] in all cases). We also observe strong agreement between our parameter estimates and those derived from alternative data sets. For example, our estimates of mutation bias and those from mutational accumulation experiments are highly correlated ([Formula: see text]). Our estimates of codon-specific translational inefficiencies and tRNA copy number-based estimates of ribosome pausing time ([Formula: see text]), and mRNA and ribosome profiling footprint-based estimates of gene expression ([Formula: see text]) are also highly correlated, thus supporting the hypothesis that selection against translational inefficiency is an important force driving the evolution of CUB. Surprisingly, we find that for particular amino acids, codon usage in highly expressed genes can still be largely driven by mutation bias and that failing to take mutation bias into account can lead to the misidentification of an amino acid's "optimal" codon. In conclusion, our method demonstrates that an enormous amount of biologically important information is encoded within genome scale patterns of codon usage, accessing this information does not require gene expression measurements, but instead carefully formulated biologically interpretable models.

Project description:BackgroundOsteosarcoma (OS) metastasis is the most common cause of cancer-related mortality, however, no sufficient clinical biomarkers have been identified. In this study, we identified five genes to help predict metastasis at diagnosis.MethodsWe performed weighted gene co-expression network analysis (WGCNA) to identify the most relevant gene modules associated with OS metastasis. An important machine learning algorithm, the support vector machine (SVM), was employed to predict key genes for classifying the OS metastasis phenotype. Finally, we investigated the clinical significance of key genes and their enriched pathways.ResultsEighteen modules were identified in WGCNA, among which the pink, red, brown, blue, and turquoise modules demonstrated good preservation. In the five modules, the brown and red modules were highly correlated with OS metastasis. Genes in the two modules closely interacted in protein-protein interaction networks and were therefore chosen for further analysis. Genes in the two modules were primarily enriched in the biological processes associated with tumorigenesis and development. Furthermore, 65 differentially expressed genes were identified as common hub genes in both WGCNA and protein-protein interaction networks. SVM classifiers with the maximum area under the curve were based on 30 and 15 genes in the brown and red modules, respectively. The clinical significance of the 45 hub genes was analyzed. Of the 45 genes, 17 were found to be significantly correlated with survival time. Finally, 5/17 genes, including ADAP2 (P = 0.0094), LCP2 (P = 0.013), ARHGAP25 (P = 0.0049), CD53 (P = 0.016), and TLR7 (P = 0.04) were significantly correlated with the metastatic phenotype. In vitro verification, western blotting, wound healing analyses, transwell invasion assays, proliferation assays, and colony formation assays indicated that ARHGAP25 promoted OS cell migration, invasion, proliferation, and epithelial-mesenchymal transition.ConclusionWe identified five genes, namely ADAP2, LCP2, ARHGAP25, CD53, and TLR7, as candidate biomarkers for the prediction of OS metastasis; ARHGAP25 inhibits MG63 OS cell growth, migration, and invasion in vitro, indicating that ARHGAP25 can serve as a promising specific and prognostic biomarker for OS metastasis.

Dataset Information

Diagnostic biases in translational bioinformatics.

Publications

Diagnostic biases in translational bioinformatics.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets