Showing papers by "Sushmita Paul published in 2022"
TL;DR: Zhang et al. as discussed by the authors reported that low activating transcription factor 2 (ATF2) levels correlated with worse prognosis and tumor aggressiveness in colorectal cancer patients.
Abstract: Abstract In cancer, the activating transcription factor 2 (ATF2) has pleiotropic functions in cellular responses to growth stimuli, damage, or inflammation. Due to only limited studies, the significance of ATF2 in colorectal cancer (CRC) is not well understood. We report that low ATF2 levels correlated with worse prognosis and tumor aggressiveness in CRC patients. NanoString gene expression and ChIP analysis confirmed trophoblast cell surface antigen 2 (TROP2) as a novel inhibitory ATF2 target gene. This inverse correlation was further observed in primary human tumor tissues. Immunostainings revealed that high intratumoral heterogeneity for ATF2 and TROP2 expression was sustained also in liver metastasis. Mechanistically, our in vitro data of CRISPR/Cas9-generated ATF2 knockout (KO) clones revealed that high TROP2 levels were critical for cell de-adhesion and increased cell migration without triggering EMT. TROP2 was enriched in filopodia and displaced Paxillin from adherens junctions. In vivo imaging, micro-computer tomography, and immunostainings verified that an ATF2 KO /TROP2 high status triggered tumor invasiveness in in vivo mouse and chicken xenograft models. In silico analysis provided direct support that ATF2 low /TROP2 high expression status defined high-risk CRC patients. Finally, our data demonstrate that ATF2 acts as a tumor suppressor by inhibiting the cancer driver TROP2. Therapeutic TROP2 targeting might prevent particularly the first steps in metastasis, i.e., the de-adhesion and invasion of colon cancer cells.
5 citations
TL;DR: In this paper , an AE-assisted cancer subtyping framework is presented that utilizes the compressed latent space of a Sparse AE neural network for multi-omics clustering, where the selected features from each of the omic data are passed to the AE.
5 citations
TL;DR: 26 algorithms/methods/tools for MRMs identification are comprehensively reviewed and they are classified into eight groups based on mathematical approaches to understand their working and suitability for one’s study.
Abstract: Abstract Identification of complex interactions between miRNAs and mRNAs in a regulatory network helps better understand the underlying biological processes. Previously, identification of these interactions was based on sequence-based predicted target binding information. With the advancement in high-throughput omics technologies, miRNA and mRNA expression for the same set of samples are available. This helps develop more efficient and flexible approaches that work by integrating miRNA and mRNA expression profiles with target binding information. Since these integrative approaches of miRNA–mRNA regulatory modules (MRMs) detection is sufficiently able to capture the minute biological details, 26 such algorithms/methods/tools for MRMs identification are comprehensively reviewed in this article. The study covers the significant features underlying every method. Therefore, the methods are classified into eight groups based on mathematical approaches to understand their working and suitability for one’s study. An algorithm could be selected based on the available information with the users and the biological question under investigation.
3 citations
15 Aug 2022
TL;DR: A novel pipeline is designed that uses a feature selection step prior to association tests to identify a crisp set of SNPs that are significantly associated with the trait under consideration and outperforms the other methods.
Abstract: Genome-wide Association Studies (GWA studies) are performed to identify genetic variants like Single Nucleotide Polymorphisms (SNPs) significantly associated with phenotype in case-control or cohort study designs. GWA studies are based on the fundamental assumption that the most statistically significant variants have a more decisive influence on the phenotype. Thus, most GWA studies use statistical approaches to identify the variants lying below a significant threshold. However, the conventional statistical techniques fail to identify significant variants for complex traits by simply thresholding since the traits are driven by both genetic and environmental factors. Therefore, it is critical to design approaches, which can capture SNPs that significantly affect the complex traits. To address this, several machine learning algorithms are being designed. However, all such techniques face the problem of a low sample to feature ratio creating redundancy and uncertainty in GWA studies. Therefore, a novel pipeline is designed that uses a feature selection step prior to association tests to identify a crisp set of SNPs that are significantly associated with the trait under consideration. The proposed pipeline combines a Rough set-based relevance technique with a machine learning-based association test called Support Vector Regression to identify cholesterol-associated SNPs. The pipeline reduces the number of SNPs to the most relevant SNPs and decreases the time required for association testing. A comparative performance analysis of the proposed approach over other existing approaches is illustrated on the pennCATH cohort dataset through R2 statistics and biological analyses. The proposed pipeline outperforms the other methods. SNP and gene enrichment studies reveal various genes, pathways and biological processes significantly related to cholesterol with the SNPs obtained from the proposed pipeline and establish the fact that the performance of the proposed rough-set-based feature selection method is significantly better.
TL;DR: In this article , a recursive integration of synergized graph representations (RISynG) is proposed to identify the most relevant feature space from each omic view and systematically integrate them.
Abstract: 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.