Classification of Tumor Samples from Expression Data Using Decision Trunks

Abstract

We present a novel machine learning approach for the classification of cancer samples using expression data. We refer to the method as “decision trunks,” since it is loosely based on decision trees, but contains several modifications designed to achieve an algorithm that: (1) produces smaller and more easily interpretable classifiers than decision trees; (2) is more robust in varying application scenarios; and (3) achieves higher classification accuracy. The decision trunk algorithm has been implemented and tested on 26 classification tasks, covering a wide range of cancer forms, experimental methods, and classification scenarios. This comprehensive evaluation indicates that the proposed algorithm performs at least as well as the current state of the art algorithms in terms of accuracy, while producing classifiers that include on average only 2–3 markers. We suggest that the resulting decision trunks have clear advantages over other classifiers due to their transparency, interpretability, and their correspondence with human decision-making and clinical testing practices.

Notch signalling mediates reproductive constraint in the adult worker honeybee

“…To identify which of the genes analysed in this study was most associated with the reproductive state of the ovary we used decision tree recursive partitioning implemented in Weka 66 . This approach has previously been used to classify tumour samples 67 68 as well as to examine the association of gene expression with ovary activity in the honeybee 20 .…”

Machine Learning for Biomarker Identification in Cancer Research – Developments Toward its Clinical Application

“…For classifying microarray datasets, hybrid of evolutionary and relative expression algorithm; an improvement over relative expression algorithm [58] iPCC Feature extraction from high throughput gene expression data; iterative employment of Pearson's correlation coefficient [53] Decision trunks For cancer gene expression data classification; novel machine-learning algorithm; an improvement over decision tree algorithm [106] SPICE System phenotype-related interplaying components enumerator for feature selection form instance-based and network-based data [107] HBSA Heuristic breadth-first search algorithm; for gene ranking based on the occurrence frequency of genes in the gene subset [108] ECD Extreme class discrimination; for feature selection, determines most discriminative variables [109] NPCA Non-negative principal component analysis; for feature-selection filter-wrapper and NPCA-support vector machine algorithm based for classification of mass spectroscopic serum proteomic patterns [110] BRL Bayesian rule learner; uses Bayesian scores and Bayesian model to build classification models [111] ellipsoidFN Ellipsoid feature net; for feature selection based on ellipsoids; analyzed on gene expression dataset [54] SVMRFE + Fisher's method A novel framework utilizing the advantages of a filtering method as well as an embedded method and redundancy reduction stage was added to address the weakness of the two. Furthermore, the proposed method uses gene ontology [112] future science group Machine learning for biomarker identification in cancer research developments toward its clinical application Review…”

Generalised Random Forest Space Overview

“…-decision trunk, proposed by Ulfenborg et al [2013], is composed of a flat series of pivot models similar to a fern, though requires the decision to be binary (say A or B), and employs pivot modules (segments) which classify into three groups, A, B meaning that it is certain that an object belongs to a respective class, and ? meaning that the decision is relayed to a next pivot classifier within the trunk.…”