Transcriptome profiling by combined machine learning and statistical R analysis identifies TMEM236 as a potential novel diagnostic biomarker for colorectal cancer

Colorectal cancer (CRC) is a common cause of cancer-related deaths worldwide. The CRC mRNA gene expression dataset containing 644 CRC tumor and 51 normal samples from the cancer genome atlas (TCGA) was pre-processed to identify the significant differentially expressed genes (DEGs). Feature selection techniques Least absolute shrinkage and selection operator (LASSO) and Relief were used along with class balancing for obtaining features (genes) of high importance. The classification of the CRC dataset was done by ML algorithms namely, random forest (RF), K-nearest neighbour (KNN), and artificial neural networks (ANN). The significant DEGs were 2933, having 1832 upregulated and 1101 downregulated genes. The CRC gene expression dataset had 23,186 features. LASSO had performed better than Relief for classifying tumor and normal samples through ML algorithms namely RF, KNN, and ANN with an accuracy of 100%, while Relief had given 79.5%, 85.05%, and 100% respectively. Common features between LASSO and DEGs were 38, from them only 5 common genes namely, VSTM2A, NR5A2, TMEM236, GDLN, and ETFDH had shown statistically significant survival analysis. Functional review and analysis of the selected genes helped in downsizing the 5 genes to 2, which are VSTM2A and TMEM236. Differential expression of TMEM236 was statistically significant and was markedly reduced in the dataset which solicits appreciation for assessment as a novel biomarker for CRC diagnosis.

Feature Extraction In Gene Expression Dataset Using Multilayer Perceptron

Numerous amount of gene expression datasets that are publicly available have accumulated since decades. It is hence essential to recognize and extract the instances in terms of quantitative and qualitative means.In this study, Keras is utilized to model the multilayer perceptron (MLP) to extract the features from the given input gene expression dataset. The MLP extracts the features from the test datasets after its initial training with the top extracted features from the training classifiers. Finally with the top extracted features, the MLP is fine tuned to extract optimal features from the gene expression datasets namely Gene Expression database of Normal and Tumor tissues 2 (GENT2). The experimental results shows that the proposed model achieves better feature selection than other methods in terms of accuracy, f-measure, precision and recall.

Optimal Predictive Model for Large Scale Classification

Biosensors calculate the expression pattern of multiple genes in experimental work. A unique genomic chip is possible to produce levels of expression from multiple genes. The ability to interpret these high-dimensional samples fuels the creation of methods of automated analysis.Even though the existing methods undergo imbalanced problems and less classification accuracy over gene expression datasets.Therefore, a novel computational method has been developed inorder to increase the classification performance of gene expression dataset and accurate disease prediction.By adding fuzzy memberships, we take into account the features of imbalanced data. Within our work, both the sample entropies and the expense for each class decide the fuzzy memberships in order to understand the different samples with various contributors to the judgment boundary. Thus, on imbalanced genomic datasets, the current proposed approach will result in more desirable classification outcomes. In addition, to build a new algorithm, we integrate the fuzzy memberships into current MKL. The results show that the proposed approach will tackle the imbalanced problem and achieve high accuracy rate.

Identification of neuroblastoma cell lines with uncommon TAZ+/mesenchymal stromal cell phenotype with strong suppressive activity on natural killer cells

BackgroundNeuroblastoma (NB) is the most common, extracranial childhood solid tumor arising from neural crest progenitor cells and is a primary cause of death in pediatric patients. In solid tumors, stromal elements recruited or generated by the cancer cells favor the development of an immune-suppressive microenvironment. Herein, we investigated in NB cell lines and in NB biopsies, the presence of cancer cells with mesenchymal phenotype and determined the immune-suppressive properties of these tumor cells on natural killer (NK) cells.MethodsWe assessed the mesenchymal stromal cell (MSC)-like phenotype and function of five human NB cell lines and the presence of this particular subset of neuroblasts in NB biopsies using flow-cytometry, immunohistochemistry, RT-qPCR, cytotoxicity assays, western blot and silencing strategy. We corroborated our data consulting a public gene-expression dataset.ResultsTwo NB cell lines, SK-N-AS and SK-N-BE(2)C, exhibited an unprecedented MSC phenotype (CD105+/CD90+/CD73+/CD29+/CD146+/GD2+/TAZ+). In these NB-MSCs, the ectoenzyme CD73 and the oncogenic/immune-regulatory transcriptional coactivator TAZ were peculiar markers. Their MSC-like nature was confirmed by their adipogenic and osteogenic differentiation potential. Immunohistochemical analysis confirmed the presence of neuroblasts with MSC phenotype (CD105+/CD73+/TAZ+). Moreover, a public gene-expression dataset revealed that, in stage IV NB, a higher expression of TAZ and CD105 strongly correlated with a poorer outcome.Among the NB-cell lines analyzed, only NB-MSCs exhibited multifactorial resistance to NK-mediated lysis, inhibition of activating NK receptors, signal adaptors and of NK-cell cytotoxicity through cell-cell contact mediated mechanisms. The latter property was controlled partially by TAZ, since its silencing in NB cells efficiently rescued NK-cell cytotoxic activity, while its overexpression induced opposite effects in non-NB-MSC cells.ConclusionsWe identified a novel NB immunoregulatory subset that: (i) displayed phenotypic and functional properties of MSC, (ii) mediated multifactorial resistance to NK-cell-induced killing and (iii) efficiently inhibited, in coculture, the cytotoxic activity of NK cells against target cells through a TAZ-dependent mechanism. These findings indicate that targeting novel cellular and molecular components may disrupt the immunomodulatory milieu of the NB microenvironment ameliorating the response to conventional treatments as well as to advanced immunotherapeutic approaches, including adoptive transfer of NK cells and chimeric antigen receptor T or NK cells.

An Improved Dragonfly Optimization Algorithm based Feature Selection in High Dimensional Gene Expression Analysis for Lung Cancer Recognition

A microarray gene expression data is an efficient dataset for analyzing expression of thousands of genes and related disease. The more accurate analysis can be obtained by comparing Gene expression of disease tissues with normal tissues which helps to recognize the type of cancer. The processing of microarray datasets such as feature selection, sampling and classification is highly challenged due to its high dimensionality. Many recent researchers used various feature selection techniques for dimensionality reduction. Dragonfly optimization Algorithm (DA) was a feature selection technique used to reduce the dimensionality of lung cancer gene expression dataset. The dragonflies in DA are flying randomly based on the model developed by using the Levy Flight Mechanism (LFM). Because of huge searching steps, LFM has some drawbacks like interruption of arbitrary flights and overflowing of the search area. In fact, DA lacks an internal resemblance that record past potential solutions that can lead to its premature convergence into local optima. So, in this paper an Improved Dragonfly optimization Algorithm (IDA) is introduced which effectively reduces the dimensionality of the lung cancer gene expression dataset. In IDA, Brownian motion method is used to solve the issues of LFM and pbest and gbest idea of Particle Swarm Optimization (PSO) is used to direct the search method for finding potential candidate solutions to further refine the search space for avoiding premature convergence. The wrapper feature selection approach is followed by IDA to select optimal subset of features. The Random Sub space (RS), Artificial Neural Network (ANN) and Sequential Minimal Optimization (SMO) classifiers are utilized for feature selection of IDA and recognize Lung cancer subtypes. The accuracy of the classifier for selected features of Dragon flies in training instances is used as fitness value of Dragon flies in each iteration. Finally, the experimental results prove the effectiveness of the IDA in terms of accuracy, precision, recall and F-measure.