Islam, M. K. | Search bioRxiv

bioRxiv2023

A Computational Approach for Structural and Functional Analyses of Disease-associated Mutations in the Human CYLD Gene

Tumor suppressor Cylindromatosis protein (CYLD) regulates NF- {kappa}B and JNK signaling pathway by cleaving K63 linked poly-ubiquitin chain from its substrate molecules and thus preventing the progression of tumorigenesis and metastasis of the cancer cells. Mutations in CYLD can cause aberrant structure and abnormal functionality leading to tumor formation. In this study, we utilized several computational tools such as PANTHER, PROVEAN, PREDICT- SNP, POLYPHEN 2, PHD SNP, PON P2, and SIFT to find out deleterious nsSNPs. We also highlighted the damaging impact of those deleterious nsSNPs on the structure and function of the CYLD utilizing Consurf, I-Mutant, SDM, Phyre2, HOPE, Swiss PDB Viewer, and Mutation 3D. We shortlisted 18 high-risk nsSNPs from a total of 446 nsSNPs recorded in the NCBI database. Based on the conservation profile, stability status, and structural impact analysis we finalized 13nsSNPs. Molecular docking analysis and molecular dynamic simulation concluded the study with the findings of two significant nsSNPs (R830K, H827R) which have a remarkable impact on binding affinity, RMSD, RMSF, Radius of gyration, and hydrogen bond formation during CYLD-ubiquitin interaction. The principal component analysis compared native and two mutants R830K, H827R of CYLD that signifies structural and energy profile fluctuations during molecular dynamic (MD) simulation. Finally, the Protein-protein interaction network showed CYLD interacts with 20 proteins involved in several biological pathways that mutations can impair. Considering all these in silico analyses, our study recommended conducting large-scale association studies of nsSNPs of CYLD with cancer as well as designing precise medications against diseases associated with these polymorphisms.

bioinformatics↗

bioRxiv2022

Deep learning-based predictive identification of functional subpopulations of hematopoietic stem cells and multipotent progenitors

Hematopoietic stem cells (HSCs) and multipotent progenitors (MPPs) are crucial for maintaining lifelong hematopoiesis. Developing methods to distinguish stem cells from other progenitors and evaluate stem cell functions has been a central task in stem cell research. Deep learning has been demonstrated as a powerful tool in cell image analysis and classification. In this study, we explored the possibility of using deep learning to differentiate HSCs and MPPs based on their light microscopy (DIC) images. After extensive training and validation with large image data sets, we successfully develop a three-class classifier (we named it the LSM model) that reliably differentiate long-term HSCs (LT-HSCs), short-term HSCs (ST-HSCs), and MPPs. Importantly, we demonstrated that our LSM model achieved its differentiating capability by learning the intrinsic morphological features from cell images. Furthermore, we showed that the performance of our LSM model was not affected by how these cells were identified and isolated, i.e., sorted by surface markers or intracellular GFP markers. Prospective identification of HSCs and MPPs in Evi1GFP transgenic mice by LSM model suggested that the cells with the highest GFP expression were LT-HSCs, and this prediction was substantiated later by a long-term competitive reconstitution assay. Moreover, based on DIC image data sets, we also successfully built another two-class classifier that can effectively distinguish aged HSCs from young HSCs, which both express the same surface markers but are functionally different. This finding is of particular interest since it may provide a novel quick and efficient approach, obviating the need for a time-consuming transplantation experiment, to evaluate the functional states of HSCs. Together, our study provides evidence for the first time that HSCs and MPPs can be differentiated by deep learning based on cell morphology. This novel and robust deep learning-based platform will provide a basis for the future development of a new generation stem cell identification and separation system. It may also provide new insight into molecular mechanisms underlying the self-renewal feature of stem cells.

bioengineering↗