A Semi-supervised Learning Approach for Pan-Cancer Somatic Genomic Variant Classification

Nicora, Giovanna; Marini, Simone; Limongelli, Ivan; Rizzo, Ettore; Montoli, S; Tricomi, Francesca Floriana; Bellazzi, Riccardo

doi:10.1007/978-3-030-21642-9_7

Cited by 5 publications

(7 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One public effort to create such a shared knowledgebase is the AACR Project GENIE, which is a multidisciplinary effort that integrates de-identified clinical-grade cancer genomic data with clinical outcome data for tens of thousands of cancer patients treated at multiple institutions worldwide (AACR Project Genie Consortium, 2017). The GENIE database (JW has grant support from GENIE) has been used to train machine-learning classifiers and is in the process of adding more in-depth phenotype information (Nicora et al, 2019). In the future, the international oncology community should continue to converge on common standards for knowledge representation, so that a comprehensive and unified knowledgebase that links tumor molecular profile data with approved therapies and available clinical trials, which would lead to the achievement of the goal of precision oncology, can be achieved.…”

Section: Discussionmentioning

confidence: 99%

A Review of Precision Oncology Knowledgebases for Determining the Clinical Actionability of Genetic Variants

Warner

2020

Front. Cell Dev. Biol.

View full text Add to dashboard Cite

The increased availability of tumor genetic testing and targeted cancer therapies contributes to the advancement of precision medicine in the field of oncology. Precision oncology knowledgebases provide a way of organizing clinically relevant genetic information in a way that is easily accessible for both oncologists and patients, facilitating the genetic-based clinical decision making. Many organizations and companies have built precision oncology knowledgebases, intended for multiple users. In general, these knowledgebases offer information on cancer-related genetic variants as well as their associated diagnostic, prognostic, and therapeutic implications, but they often differ in their information curations, designs, and user experiences. It is advisable that oncologists use multiple knowledgebases during their practice to have them complement each other. In the future, convergence toward common standards and formats is needed to ensure that the comprehensive knowledge across all sources can be unified to bring the oncology community closer to the achievement of the goal of precision oncology.

show abstract

Section: Discussionmentioning

confidence: 99%

A Review of Precision Oncology Knowledgebases for Determining the Clinical Actionability of Genetic Variants

Warner

2020

Front. Cell Dev. Biol.

View full text Add to dashboard Cite

show abstract

“…The first proposals by Carter et al [19] and Capriotti et al [20] were based on these algorithms. Among the SVM-based approaches, whereas most papers adopted the traditional SVM algorithm [20,22,24,27,31,32,39,55,56,57,58], we observed three papers using OneClass SVM [45,49,59] and one paper using Sequential Minimal Optimization (SMO) [28]. SVM is a popular and consolidated technique in the field, as it continues to be largely applied throughout the years since 2011.…”

Section: Methods Based On Supervised Learningmentioning

confidence: 99%

“…Ensemble methods train multiple weak classifiers, such as decision trees, and combine their output (e.g., with majority voting) to achieve a better predictive performance. We observed a frequent use of Random Forests (RF) [19,26,27,34,37,38,43,49,53,54,55,56,57,58], as well as Gradient Boosting Trees (GBT) [37,42], and eXtreme Gradient Boosting (XGBoost) [57,58]. While RF [114] uses bagging (i.e., bootstrap aggregating) and random features subsets to train multiple and diverse trees independently, Gradient Boosting [115] builds one tree at a time, introducing a weak learner to improve shortcoming of existing trees by assigning more weight on instances with wrong predictions and high errors.…”

Section: Methods Based On Supervised Learningmentioning

confidence: 99%

“…Regression-based methods appeared in 11 selected papers, most of which adopted logistic regression [26,29,31,32,37,56,57]. We also found papers using regularized regressions, including Ridge [49] and Lasso regression [23]. Regularization aims to discourage complex models by penalizing the magnitude of the coefficients and the error term.…”

Section: Methods Based On Supervised Learningmentioning

confidence: 99%

“…The low number of related papers published in 2021 is justified by the short period covered by our analysis. Papers were published mainly in scientific journals, with only five [21,36,49,50,55] appearing in conference proceedings. The top three journals in terms of number of papers were Bioinformatics (6), Journal of Computational Biology (3), and Plos One (3).…”

Section: Overview Of Selected Papersmentioning

confidence: 99%

See 2 more Smart Citations

Machine learning methods for prediction of cancer driver genes: a survey paper

Andrades,

Recamonde-Mendoza

2021

Preprint

View full text Add to dashboard Cite

Identifying the genes and mutations that drive the emergence of tumors is a major step to improve understanding of cancer and identify new directions for disease diagnosis and treatment. Despite the large volume of genomics data, the precise detection of driver mutations and their carrying genes, known as cancer driver genes, from the millions of possible somatic mutations remains a challenge. Computational methods play an increasingly important role in identifying genomic patterns associated with cancer drivers and developing models to predict driver events. Machine learning (ML) has been the engine behind many of these efforts and provides excellent opportunities for tackling remaining gaps in the field. Thus, this survey aims to perform a comprehensive analysis of ML-based computational approaches to identify cancer driver mutations and genes, providing an integrated, panoramic view of the broad data and algorithmic landscape within this scientific problem. We discuss how the interactions among data types and ML algorithms have been explored in previous solutions and outline current analytical limitations that deserve further attention from the scientific community. We hope that by helping readers become more familiar with significant developments in the field brought by ML, we may inspire new researchers to address open problems and advance our knowledge towards cancer driver discovery.

show abstract