Data Mining - Concepts and Techniques.

The hallmarks of cancer conceptualization is a heuristic tool for distilling the vast complexity of cancer phenotypes and genotypes into a provisional set of underlying principles. As knowledge of cancer mechanisms has progressed, other facets of the disease have emerged as potential refinements. Herein, the prospect is raised that phenotypic plasticity and disrupted differentiation is a discrete hallmark capability, and that nonmutational epigenetic reprogramming and polymorphic microbiomes both constitute distinctive enabling characteristics that facilitate the acquisition of hallmark capabilities. Additionally, senescent cells, of varying origins, may be added to the roster of functionally important cell types in the tumor microenvironment. SIGNIFICANCE: Cancer is daunting in the breadth and scope of its diversity, spanning genetics, cell and tissue biology, pathology, and response to therapy. Ever more powerful experimental and computational tools and technologies are providing an avalanche of "big data" about the myriad manifestations of the diseases that cancer encompasses. The integrative concept embodied in the hallmarks of cancer is helping to distill this complexity into an increasingly logical science, and the provisional new dimensions presented in this perspective may add value to that endeavor, to more fully understand mechanisms of cancer development and malignant progression, and apply that knowledge to cancer medicine. 

/pdf/hallmarks-of-cancer-new-dimensions-fgechgl0.pdf

Hallmarks of Cancer: New Dimensions

50 Years of Image Analysis

The hallmarks of cancer conceptualization is a heuristic tool for distilling the vast complexity of cancer phenotypes and genotypes into a provisional set of underlying principles. As knowledge of cancer mechanisms has progressed, other facets of the disease have emerged as potential refinements. Herein, the prospect is raised that phenotypic plasticity and disrupted differentiation is a discrete hallmark capability, and that nonmutational epigenetic reprogramming and polymorphic microbiomes both constitute distinctive enabling characteristics that facilitate the acquisition of hallmark capabilities. Additionally, senescent cells, of varying origins, may be added to the roster of functionally important cell types in the tumor microenvironment. SIGNIFICANCE: Cancer is daunting in the breadth and scope of its diversity, spanning genetics, cell and tissue biology, pathology, and response to therapy. Ever more powerful experimental and computational tools and technologies are providing an avalanche of "big data" about the myriad manifestations of the diseases that cancer encompasses. The integrative concept embodied in the hallmarks of cancer is helping to distill this complexity into an increasingly logical science, and the provisional new dimensions presented in this perspective may add value to that endeavor, to more fully understand mechanisms of cancer development and malignant progression, and apply that knowledge to cancer medicine.

Hallmarks of Cancer: New Dimensions.

From bench to bedside

Selection of reliable genes from a huge gene expression data containing high intergene correlation is essential to carry out a diagnostic test and successful treatment. In this regard, a rough set based gene selection algorithm is reported, which selects a set of genes by maximizing the relevance and significance of the selected genes. A gene ontology-based similarity measure is proposed to analyze the functional diversity of the selected genes. It also helps to analyze the effectiveness of different gene selection methods. The performance of the rough set based gene selection algorithm, along with a comparison with other gene selection methods, is studied using the predictive accuracy of K-nearest neighbor rule and support vector machine on two cancer and one arthritis microarray data sets. An important finding is that the rough set based gene selection algorithm selects more functionally diverse set of genes than the existing algorithms.

Rough sets for selection of functionally diverse genes from microarray data

Abstract Cancer subtypes identification is one of the critical steps toward advancing personalized anti-cancerous therapies. Accumulation of a massive amount of multi-platform omics data measured across the same set of samples provides an opportunity to look into this deadly disease from several views simultaneously. Few integrative clustering approaches are developed to capture shared information from all the views to identify cancer subtypes. However, they have certain limitations. The challenge here is identifying the most relevant feature space from each omic view and systematically integrating them. Both the steps should lead toward a global clustering solution with biological significance. In this respect, a novel multi-omics clustering algorithm named RISynG (Recursive Integration of Synergised Graph-representations) is presented in this study. RISynG represents each omic view as two representation matrices that are Gramian and Laplacian. A parameterised combination function is defined to obtain a synergy matrix from these representation matrices. Then a recursive multi-kernel approach is applied to integrate the most relevant, shared, and complementary information captured via the respective synergy matrices. At last, clustering is applied to the integrated subspace. RISynG is benchmarked on five multi-omics cancer datasets taken from The Cancer Genome Atlas. The experimental results demonstrate RISynG’s efficiency over the other approaches in this domain. 

/pdf/recursive-integration-of-synergised-graph-representations-of-2cij4sk4.pdf

Recursive integration of synergised graph representations of multi-omics data for cancer subtypes identification

The advent of DNA microarray technologies has revolutionized the experimental study of gene expression Microarrays have been used to study different kinds of biological processes

Possibilistic Biclustering for Discovering Value-Coherent Overlapping $$\delta $$-Biclusters

Cancer subtypes identification is very important for the advancement of precision cancer disease diagnosis and therapy. It is one of the important components of the personalized medicine framework. Cervical cancer (CC) is one of the leading gynecological cancers that causes deaths in women worldwide. However, there is a lack of studies to identify histological subtypes among the patients suffering from tumor of the uterine cervix. Hence, sub-typing of cancer can help in analyzing shared molecular profiles between different histological subtypes of solid tumors of uterine cervix. With the advancement in technology, large scale multi-omics data are generated. The integration of genomics data generated from different platforms helps in capturing complementary information about the patients. Several computational approaches have been developed that integrate muti-omics data for cancer sub-typing. In this study, mRNA (messenger RNA) and miRNA (microRNA) expression data are integrated to identify the histological subtypes of CC. In this regard, a method is proposed that ranks the biomarkers (mRNA and miRNA) on the basis of their varying expression across the samples. The ranking method generates a weight for every biomarker which is further used to identify the similarity between the samples. A well-known approach named Similarity Network Fusion (SNF) is then applied, followed by Spectral clustering, to identify groups of related samples. This study focuses on the role of weighing the biomarkers prior to their integration and application of the clustering algorithm. The weighing method proposed in this study is compared with some other methods and proved to be more efficient. The proposed method helps in identifying histological subtypes of CC and can also be applied to other types of cancer data where histological subtypes play a key role in designing treatments and therapies.

Importance of Feature Weighing in Cervical Cancer Subtypes Identification

Clustering is used to identify natural groups present in the data. It has been applied widely for analyzing gene expression data to discover gene clusters that might be involved in same biological processes. This information is very important for analyzing data of fatal diseases like cancers and identifying potential diagnostic and prognostic markers. Existing clustering methods used in this regard are computationally efficient, but do not always produce biologically meaningful results. Additionally, they have one or the other shortcomings; either they are not able to deal with arbitrary-shaped clusters, require number of clusters to be specified previously or are not efficient in dealing with noise present in biological data. In this study, a new density-based clustering method specific for gene expression data is introduced that overcomes the above shortcomings and produces biologically enriched clusters of functionally similar genes by incorporating biological information from Gene Ontology (GO). The proposed method integrates the GO semantic similarity information and the correlation information between the genes for obtaining clusters. The clusters are further validated for their biological relevance using Disease Ontology, KEGG Pathway enrichment and protein-protein interaction network analysis.

Sushmita Paul

Papers

Rough sets for selection of functionally diverse genes from microarray data

Recursive integration of synergised graph representations of multi-omics data for cancer subtypes identification

Possibilistic Biclustering for Discovering Value-Coherent Overlapping $$\delta $$-Biclusters

Importance of Feature Weighing in Cervical Cancer Subtypes Identification

Density-Based Clustering of Functionally Similar Genes Using Biological Knowledge