Applications of Support Vector Machine (SVM) Learning in Cancer Genomics

Cancer Genomics & Proteomics, Journal Year: 2018, Volume and Issue: 15(1)

Published: Jan. 2, 2018

Machine learning with maximization (support) of separating margin (vector), called support vector machine (SVM) learning, is a powerful classification tool that has been used for cancer genomic or subtyping. Today, as advancements in high-throughput technologies lead to production large amounts and epigenomic data, the feature SVMs expanding its use genomics, leading discovery new biomarkers, drug targets, better understanding driver genes. Herein we reviewed recent progress studies. We intend comprehend strength SVM future perspective applications.

Language: Английский

---

Big Data Analysis Using Modern Statistical and Machine Learning Methods in Medicine

International Neurourology Journal, Journal Year: 2014, Volume and Issue: 18(2), P. 50 - 50

Published: Jan. 1, 2014

In this article we introduce modern statistical machine learning and bioinformatics approaches that have been used in relationships from big data medicine behavioral science typically include clinical, genomic (and proteomic) environmental variables. Every year, collected biomedical is getting larger more complicated. Thus, medicine, also need to be aware of trend understand the tools are available analyze these datasets. Many analyses aimed such datasets introduced recently. However, given many different types genomic, data, it rather uncommon see methods combine knowledge resulting those types. To extent, will terms clinical single nucleotide polymorphism gene expression studies their interactions with environment. article, concept well-known regression as linear logistic regressions has widely models Bayesian networks complicated data. Also discuss how represent interaction among using models. We conclude a promising method called suitable analyzing sets consists type large Such model form provide us comprehensive understanding human physiology disease. Keywords: analysis; Statistical interpretation; Systems biology

Language: Английский

---

Machine learning approaches for the discovery of gene-gene interactions in disease data

Briefings in Bioinformatics, Journal Year: 2012, Volume and Issue: 14(2), P. 251 - 260

Published: May 18, 2012

Because of the complexity gene–phenotype relationships machine learning approaches have considerable appeal as a strategy for modelling interactions. A number such methods been developed and applied in recent years with some modest success. Progress is hampered by challenges presented disease genetic data, including phenotypic heterogeneity, polygenic forms inheritance variable penetrance, combined analytical computational issues arising from enormous potential We review here current focusing, wherever possible, on applications to real data (particularly context genome-wide association studies) looking ahead further posed next generation sequencing data.

Language: Английский

---

The Utility of Independent Component Analysis and Machine Learning in the Identification of the Amyotrophic Lateral Sclerosis Diseased Brain

Frontiers in Human Neuroscience, Journal Year: 2013, Volume and Issue: 7

Published: Jan. 1, 2013

Amyotrophic lateral sclerosis (ALS) is a devastating disease with lifetime risk of ∼1 in 2000. Presently, diagnosis ALS relies on clinical assessments for upper motor neuron and lower deficits multiple body segments together history progression symptoms. In addition, it common to evaluate pathology by electromyography. However, solely assessed grounds, thus hindering diagnosis. the past decade magnetic resonance methods have been shown be sensitive process, namely: resting-state connectivity measured functional MRI, cortical thickness high-resolution imaging, diffusion tensor imaging (DTI) metrics such as fractional anisotropy radial diffusivity, more recently spectroscopy (MRS) measures gamma-aminobutyric acid concentration. this present work we utilize independent component analysis derive brain networks based use those derived build state classifier using machine learning (support-vector machine). We show that possible achieve over 71% accuracy classification. These results are promising development clinically relevant classifier. Future inclusion other MR modalities structural DTI MRS should improve overall accuracy.

Language: Английский

---

Development of an Automated MRI-Based Diagnostic Protocol for Amyotrophic Lateral Sclerosis Using Disease-Specific Pathognomonic Features: A Quantitative Disease-State Classification Study

PLoS ONE, Journal Year: 2016, Volume and Issue: 11(12), P. e0167331 - e0167331

Published: Dec. 1, 2016

Background Despite significant advances in quantitative neuroimaging, the diagnosis of ALS remains clinical and MRI-based biomarkers are not currently used to aid diagnosis. The objective this study is develop a robust, disease-specific, multimodal classification protocol validate its diagnostic accuracy independent, early-stage follow-up data sets. Methods 147 participants (81 patients 66 healthy controls) were divided into training sample validation sample. Patients underwent imaging longitudinally. After removing age-related variability, indices grey white matter integrity ALS-specific pathognomonic brain regions included cross-validated binary logistic regression model determine probability individual scans indicating ALS. following anatomical assessed for classification: average density left right precentral gyrus, fractional anisotropy radial diffusivity superior corona radiata, inferior internal capsule, mesencephalic crus cerebral peduncles, pontine segment corticospinal tract, values genu, corpus splenium callosum. Results Using 50% cut-off value suffering from ALS, was able discriminate HC with good sensitivity (80.0%) moderate (70.0%) (85.7%) (78.4%) independent Conclusions This endeavours advance biomarker research towards pragmatic applications by providing an approach automated individual-data interpretation based on group-level observations.

Language: Английский

---

The segmental diffusivity profile of amyotrophic lateral sclerosis associated white matter degeneration

European Journal of Neurology, Journal Year: 2016, Volume and Issue: 23(8), P. 1361 - 1371

Published: May 21, 2016

Background and purpose Magnetic resonance diffusivity indices have been repeatedly proposed as biomarkers of neurodegeneration in amyotrophic lateral sclerosis ( ALS ), but no consensus exists to which parameter is the most sensitive identify early degenerative changes. Despite numerous studies, surprisingly little known segmental vulnerability corticospinal tracts corpus callosum. Our objective was characterize core three‐dimensional white matter signature , describe phenotype‐specific patterns degeneration evaluate profile individual patients controls specific segments. Methods A large neuroimaging study undertaken with 62 55 age‐matched healthy controls. White alterations were explored based on fractional anisotropy radial, mean axial indices. Atlas‐based region interest analyses carried out corona radiata, internal capsules, cerebral peduncles, splenium, body genu Percentage change receiver operating characteristic ROC ) curves used disease‐state discriminating measures regions. Results Bulbar onset exhibit extensive corticobulbar tract involvement capsule fibres radiata subjacent bulbar representation motor homunculus. Spinal show predominantly posterior medial pathology. curve revealed that crura best discriminate (area under 80.1%). Conclusions Amyotrophic associated a core, disease‐specific demonstrated by radial measurements. The main phenotypes are manifestations relatively selective fibres.

Language: Английский

---

A Markov blanket-based method for detecting causal SNPs in GWAS

BMC Bioinformatics, Journal Year: 2010, Volume and Issue: 11(S3)

Published: April 1, 2010

Detecting epistatic interactions associated with complex and common diseases can help to improve prevention, diagnosis treatment of these diseases. With the development genome-wide association studies (GWAS), designing powerful robust computational method for identifying becomes a great challenge bioinformatics society, because study often deals large size genotyped data huge amount combinations all possible genetic factors. Most existing detection methods are based on classification capacity SNP sets, which may fail identify sets that strongly introduce lot false positives. In addition, most not suitable scale due their complexity. We propose new Markov Blanket-based method, DASSO-MB (Detection ASSOciations using Blanket) detect in case-control GWAS. blanket target variable T completely shield from other variables. Thus, we guarantee set detected by has strong contains fewest Furthermore, uses heuristic search strategy calculating between variables avoid time-consuming training process as machine-learning methods. apply our algorithm simulated datasets real dataset. compare commonly-used show significantly outperforms is capable finding SNPs Our shows minimal causal diseases, less positives compared Given genomic dataset produced GWAS, this critical saving potential costs biological experiments being an efficient guideline pathogenesis research.

Language: Английский

---

A Review for Detecting Gene-Gene Interactions Using Machine Learning Methods in Genetic Epidemiology

BioMed Research International, Journal Year: 2013, Volume and Issue: 2013, P. 1 - 13

Published: Jan. 1, 2013

Recently, the greatest statistical computational challenge in genetic epidemiology is to identify and characterize genes that interact with other environment factors bring effect on complex multifactorial disease. These gene-gene interactions are also denoted as epitasis which this phenomenon cannot be solved by traditional method due high dimensionality of data occurrence multiple polymorphism. Hence, there several machine learning methods solve such problems identifying susceptibility gene neural networks (NNs), support vector (SVM), random forests (RFs) common This paper gives an overview methods, describing methodology each its application detecting gene-environment interactions. Lastly, discussed presents strengths weaknesses human

Language: Английский

---

Genetic studies of complex human diseases: Characterizing SNP-disease associations using Bayesian networks

BMC Systems Biology, Journal Year: 2012, Volume and Issue: 6(S3)

Published: Dec. 1, 2012

Abstract Background Detecting epistatic interactions plays a significant role in improving pathogenesis, prevention, diagnosis, and treatment of complex human diseases. Applying machine learning or statistical methods to interaction detection will encounter some common problems, e.g., very limited number samples, an extremely high search space, large false positives, ways measure the association between disease markers phenotype. Results To address problems computational detection, we propose score-based Bayesian network structure method, EpiBN, detect interactions. We apply proposed method both simulated datasets three real datasets. Experimental results on simulation data show that our outperforms other commonly-used terms power sample-efficiency, is especially suitable for detecting with weak no marginal effects. Furthermore, scalable data. Conclusions network-based In develop new scoring function, which can reflect higher-order by estimating model complexity from data, fast Branch-and-Bound algorithm learn two-layer containing only one target node. make use Markov chain Monte Carlo (MCMC) perform screening process. Applications GWAS (genome-wide studies) may provide helpful insights into understanding genetic basis Age-related Macular Degeneration, late-onset Alzheimer's disease, autism.

Language: Английский

---

The selective anatomical vulnerability of ALS: ‘disease-defining’ and ‘disease-defying’ brain regions

Amyotrophic Lateral Sclerosis and Frontotemporal Degeneration, Journal Year: 2016, Volume and Issue: 17(7-8), P. 561 - 570

Published: April 18, 2016

A large multiparametric MRI study has been undertaken to evaluate anatomical patterns of basal ganglia, white matter and cortical grey involvement in ALS. Unaffected brain regions are mapped patients with significant disability. Multiple diffusivity measures, density alterations, ganglia volumes subcortical atrophy evaluated. Results demonstrated a strikingly selective vulnerability pattern ALS that preferentially affects specific structures, commissural tracts regions, suggestive networkwise neurodegeneration In conclusion, pathology exhibits predilection for inter-connected sites can be comprehensively characterized vivo by neuroimaging. The systematic characterization unaffected implications the development classifier analyses elucidation disease biology. sparing contiguous raises important pathophysiological, phylogenetic ontogenetic questions regarding pathogenesis spread.

Language: Английский

---