CuMiDa: An Extensively Curated Microarray Database for Benchmarking and Testing of Machine Learning Approaches in Cancer Research

Gene Expression-Based Cancer Classification for Handling the Class Imbalance Problem and Curse of Dimensionality

Cancer is a leading cause of death globally. The majority of cancer cases are only diagnosed in the late stages of cancer due to the use of conventional methods. This reduces the chance of survival for cancer patients. Therefore, early detection consequently followed by early diagnoses are important tasks in cancer research. Gene expression microarray technology has been applied to detect and diagnose most types of cancers in their early stages and has gained encouraging results. In this paper, we address the problem of classifying cancer based on gene expression for handling the class imbalance problem and the curse of dimensionality. The oversampling technique is utilized to overcome this problem by adding synthetic samples. Another common issue related to the gene expression dataset addressed in this paper is the curse of dimensionality. This problem is addressed by applying chi-square and information gain feature selection techniques. After applying these techniques individually, we proposed a method to select the most significant genes by combining those two techniques (CHiS and IG). We investigated the effect of these techniques individually and in combination. Four benchmarking biomedical datasets (Leukemia-subtypes, Leukemia-ALLAML, Colon, and CuMiDa) were used. The experimental results reveal that the oversampling techniques improve the results in most cases. Additionally, the performance of the proposed feature selection technique outperforms individual techniques in nearly all cases. In addition, this study provides an empirical study for evaluating several oversampling techniques along with ensemble-based learning. The experimental results also reveal that SVM-SMOTE, along with the random forests classifier, achieved the highest results, with a reporting accuracy of 100%. The obtained results surpass the findings in the existing literature as well.

Linear programming based computational technique for leukemia classification using gene expression profile

Cancer is a serious public health concern worldwide and is the leading cause of death. Blood cancer is one of the most dangerous types of cancer. Leukemia is a type of cancer that affects the blood cell and bone marrow. Acute leukemia is a chronic condition that is fatal if left untreated. A timely, reliable, and accurate diagnosis of leukemia at an early stage is critical to treating and preserving patients' lives. There are four types of leukemia, namely acute lymphocytic leukemia, acute myelogenous leukemia, chronic lymphocytic in extracting, and chronic myelogenous leukemia. Recognizing these cancerous development cells is often done via manual analysis of microscopic images. This requires an extraordinarily skilled pathologist. Leukemia symptoms might include lethargy, a lack of energy, a pale complexion, recurrent infections, and easy bleeding or bruising. One of the challenges in this area is identifying subtypes of leukemia for specialized treatment. This Study is carried out to increase the precision of diagnosis to assist in the development of personalized plans for treatment, and improve general leukemia-related healthcare practises. In this research, we used leukemia gene expression data from Curated Microarray Database (CuMiDa). Microarrays are ideal for studying cancer, however, categorizing the expression pattern of microarray information can be challenging. This proposed study uses feature selection methods and machine learning techniques to predict and classify subtypes of leukemia in gene expression data CuMiDa (GSE9476). This research work utilized linear programming (LP) as a machine-learning technique for classification. Linear programming model classifies and predicts the subtypes of leukemia Bone_Marrow_CD34, Bone Marrow, AML, PB, and PBSC CD34. Before using the LP model, we selected 25 features from the given dataset of 22283 features. These 25 significant features were the most distinguishing for classification. The classification accuracy of this work is 98.44%.

Development of Symbolic Expressions Ensemble for Breast Cancer Type Classification Using Genetic Programming Symbolic Classifier and Decision Tree Classifier

Breast cancer is a type of cancer with several sub-types. It occurs when cells in breast tissue grow out of control. The accurate sub-type classification of a patient diagnosed with breast cancer is mandatory for the application of proper treatment. Breast cancer classification based on gene expression is challenging even for artificial intelligence (AI) due to the large number of gene expressions. The idea in this paper is to utilize the genetic programming symbolic classifier (GPSC) on the publicly available dataset to obtain a set of symbolic expressions (SEs) that can classify the breast cancer sub-type using gene expressions with high classification accuracy. The initial problem with the used dataset is a large number of input variables (54,676 gene expressions), a small number of dataset samples (151 samples), and six classes of breast cancer sub-types that are highly imbalanced. The large number of input variables is solved with principal component analysis (PCA), while the small number of samples and the large imbalance between class samples are solved with the application of different oversampling methods generating different dataset variations. On each oversampled dataset, the GPSC with random hyperparameter values search (RHVS) method is trained using 5-fold cross validation (5CV) to obtain a set of SEs. The best set of SEs is chosen based on mean values of accuracy (ACC), the area under the receiving operating characteristic curve (AUC), precision, recall, and F1-score values. In this case, the highest classification accuracy is equal to 0.992 across all evaluation metric methods. The best set of SEs is additionally combined with a decision tree classifier, which slightly improves ACC to 0.994.

A Hybrid Machine Learning Approach to Screen Optimal Predictors for the Classification of Primary Breast Tumors from Gene Expression Microarray Data

The high dimensionality and sparsity of the microarray gene expression data make it challenging to analyze and screen the optimal subset of genes as predictors of breast cancer (BC). The authors in the present study propose a novel hybrid Feature Selection (FS) sequential framework involving minimum Redundancy-Maximum Relevance (mRMR), a two-tailed unpaired t-test, and meta-heuristics to screen the most optimal set of gene biomarkers as predictors for BC. The proposed framework identified a set of three most optimal gene biomarkers, namely, MAPK 1, APOBEC3B, and ENAH. In addition, the state-of-the-art supervised Machine Learning (ML) algorithms, namely Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Neural Net (NN), Naïve Bayes (NB), Decision Tree (DT), eXtreme Gradient Boosting (XGBoost), and Logistic Regression (LR) were used to test the predictive capability of the selected gene biomarkers and select the most effective breast cancer diagnostic model with higher values of performance matrices. Our study found that the XGBoost-based model was the superior performer with an accuracy of 0.976 ± 0.027, an F1-Score of 0.974 ± 0.030, and an AUC value of 0.961 ± 0.035 when tested on an independent test dataset. The screened gene biomarkers-based classification system efficiently detects primary breast tumors from normal breast samples.

Constrained neuro fuzzy inference methodology for explainable personalised modelling with applications on gene expression data

Interpretable machine learning models for gene expression datasets are important for understanding the decision-making process of a classifier and gaining insights on the underlying molecular processes of genetic conditions. Interpretable models can potentially support early diagnosis before full disease manifestation. This is particularly important yet, challenging for mental health. We hypothesise this is due to extreme heterogeneity issues which may be overcome and explained by personalised modelling techniques. Thus far, most machine learning methods applied to gene expression datasets, including deep neural networks, lack personalised interpretability. This paper proposes a new methodology named personalised constrained neuro fuzzy inference (PCNFI) for learning personalised rules from high dimensional datasets which are structurally and semantically interpretable. Case studies on two mental health related datasets (schizophrenia and bipolar disorders) have shown that the relatively short and simple personalised fuzzy rules provided enhanced interpretability as well as better classification performance compared to other commonly used machine learning methods. Performance test on a cancer dataset also showed that PCNFI matches previous benchmarks. Insights from our approach also indicated the importance of two genes (ATRX and TSPAN2) as possible biomarkers for early differentiation of ultra-high risk, bipolar and healthy individuals. These genes are linked to cognitive ability and impulsive behaviour. Our findings suggest a significant starting point for further research into the biological role of cognitive and impulsivity-related differences. With potential applications across bio-medical research, the proposed PCNFI method is promising for diagnosis, prognosis, and the design of personalised treatment plans for better outcomes in the future.

Synthetic lethal gene pairs: Experimental approaches and predictive models

Synthetic lethality (SL) refers to a genetic interaction in which the simultaneous perturbation of two genes leads to cell or organism death, whereas viability is maintained when only one of the pair is altered. The experimental exploration of these pairs and predictive modeling in computational biology contribute to our understanding of cancer biology and the development of cancer therapies. We extensively reviewed experimental technologies, public data sources, and predictive models in the study of synthetic lethal gene pairs and herein detail biological assumptions, experimental data, statistical models, and computational schemes of various predictive models, speculate regarding their influence on individual sample- and population-based synthetic lethal interactions, discuss the pros and cons of existing SL data and models, and highlight potential research directions in SL discovery.

Predicting prediction: A systematic workflow to analyze factors affecting the classification performance in genomic biomarker discovery

High throughput technologies in genomics enable the analysis of small alterations in gene expression levels. Patterns of such deviations are an important starting point for the discovery and verification of new biomarker candidates. Identifying such patterns is a challenging task that requires sophisticated machine learning approaches. Currently, there are a variety of classification models, and a common approach is to compare the performance and select the best one for a given classification problem. Since the association between the features of a data set and the performance of a particular classification method is still not fully understood, the main contribution of this work is to provide a new methodology for predicting the prediction results of different classifiers in the field of biomarker discovery. We propose here a three-steps computational workflow that includes an analysis of the data set characteristics, the calculation of the classification accuracy and, finally, the prediction of the resulting classification error. The experiments were carried out on synthetic and microarray datasets. Using this method, we showed that the predictability strongly depends on the discriminatory ability of the features, e.g., sets of genes, in two or multi-class datasets. If a dataset has a certain discriminatory ability, this method enables prediction of the classification performance before applying a learning model. Thus, our results contribute to a better understanding of the relationship between dataset characteristics and the corresponding performance of a machine learning method, and suggest the optimal classification method for a given dataset based on its discriminatory ability.

Comparison of machine learning techniques to handle imbalanced COVID-19 CBC datasets

The Coronavirus pandemic caused by the novel SARS-CoV-2 has significantly impacted human health and the economy, especially in countries struggling with financial resources for medical testing and treatment, such as Brazil's case, the third most affected country by the pandemic. In this scenario, machine learning techniques have been heavily employed to analyze different types of medical data, and aid decision making, offering a low-cost alternative. Due to the urgency to fight the pandemic, a massive amount of works are applying machine learning approaches to clinical data, including complete blood count (CBC) tests, which are among the most widely available medical tests. In this work, we review the most employed machine learning classifiers for CBC data, together with popular sampling methods to deal with the class imbalance. Additionally, we describe and critically analyze three publicly available Brazilian COVID-19 CBC datasets and evaluate the performance of eight classifiers and five sampling techniques on the selected datasets. Our work provides a panorama of which classifier and sampling methods provide the best results for different relevant metrics and discuss their impact on future analyses. The metrics and algorithms are introduced in a way to aid newcomers to the field. Finally, the panorama discussed here can significantly benefit the comparison of the results of new ML algorithms.

Benchmarking and Testing Machine Learning Approaches with BARRA:CuRDa, a Curated RNA-Seq Database for Cancer Research

RNA-seq is gradually becoming the dominating technique employed to access the global gene expression in biological samples, allowing more flexible protocols and robust analysis. However, the nature of RNA-seq results imposes new data-handling challenges when it comes to computational analysis. With the increasing employment of machine learning (ML) techniques in biomedical sciences, databases that could provide curated data sets treated with state-of-the-art approaches already adapted to ML protocols, become essential for testing new algorithms. In this study, we present the Benchmarking of ARtificial intelligence Research: Curated RNA-seq Database (BARRA:CuRDa). BARRA:CuRDa was built exclusively for cancer research and is composed of 17 handpicked RNA-seq data sets for Homo sapiens that were gathered from the Gene Expression Omnibus, using rigorous filtering criteria. All data sets were individually submitted to sample quality analysis, removal of low-quality bases and artifacts from the experimental process, removal of ribosomal RNA, and estimation of transcript-level abundance. Moreover, all data sets were tested using standard approaches in the field, which allows them to be used as benchmark to new ML approaches. A feature selection analysis was also performed on each data set to investigate the biological accuracy of basic techniques. Results include genes already related to their specific tumoral tissue a large amount of long noncoding RNA and pseudogenes. BARRA:CuRDa is available at http://sbcb.inf.ufrgs.br/barracurda.

Multi-Approach Bioinformatics Analysis of Curated Omics Data Provides a Gene Expression Panorama for Multiple Cancer Types

Studies describing the expression patterns and biomarkers for the tumoral process increase in number every year. The availability of new datasets, although essential, also creates a confusing landscape where common or critical mechanisms are obscured amidst the divergent and heterogeneous nature of such results. In this work, we manually curated the Gene Expression Omnibus using rigorous filtering criteria to select the most homogeneous and highest quality microarray and RNA-seq datasets from multiple types of cancer. By applying systems biology approaches, combined with machine learning analysis, we investigated possible frequently deregulated molecular mechanisms underlying the tumoral process. Our multi-approach analysis of 99 curated datasets, composed of 5,406 samples, revealed 47 differentially expressed genes in all analyzed cancer types, which were all in agreement with the validation using TCGA data. Results suggest that the tumoral process is more related to the overexpression of core deregulated machinery than the underexpression of a given gene set. Additionally, we identified gene expression similarities between different cancer types not described before and performed an overall survival analysis using 20 cancer types. Finally, we were able to suggest a core regulatory mechanism that could be frequently deregulated.