Biology subjects

Dsouza, N. R.

Publications and source records attributed to Dsouza, N. R..

3 recordsLinked to original sources

bioRxiv2023

Assessing Protein Surface-Based Scoring for Interpreting Genomic Variants

Clinical genomics sequencing is rapidly expanding the number of variants that need to be functionally elucidated. Interpreting genetic variants (i.e., mutations) usually begins by identifying how they affect protein-coding sequences. Still, the 3D protein molecule is rarely considered for large-scale variant analysis, nor for their effects on how proteins interact with each other and their environment. We propose a standardized approach to scoring protein surface property changes as a new dimension for functionally and mechanistically interpreting genomic variants. Further, it directs hypothesis generation for functional genomics research to learn more about the encoded proteins function. We developed a novel method leveraging 3D structures and time-dependent simulations to score and statistically evaluate protein surface property changes. We evaluated positive controls composed of 8 thermophilic versus mesophilic orthologs and variants that experimentally change protein solubility, which all showed large and statistically significant differences in charge distribution (p<0.01). We scored static 3D structures and dynamic ensembles for 43 independent variants (23 pathogenic and 20 uninterpreted) across four proteins. Focusing on the potassium ion channel, KCNK9, average local surface potential shifts were 0.41 kBT/ec with an average p-value of 1x10-2. In contrast, dynamic ensemble shifts averaged 1.15 kBT/ec and an average p-value of 1x10-5, enabling the identification of changes far from mutated sites. This study demonstrates that an objective assessment of how mutations affect electrostatic distributions of protein surfaces can aid in interpreting genomic variants discovered through clinical genomics sequencing.

bioinformatics↗

bioRxiv2019

Integration of Multi-level Molecular Scoring for the Interpretation of RAS-Family Genetic Variation

Protein-coding genetic variants are the first considered in most studies and Precision Medicine workflows, but their interpretation is primarily driven by DNA sequence-based analytical tools and annotations. Thus, more specific and mechanistic interpretations should be attainable by integrating DNA-based scores with scores from the protein 3D structure. However, reliable and reproducible standardization of methods that use 3D structure for genomic variation is still lacking. Further, we believe that the current paradigm of aiming to directly predict the pathogenicity of variants skips the critical step of inferring, with precision, molecular mechanisms of dysfunction. Thus, we report herein the development and evaluation of single and composite 3D structure-based scores and their integration with protein and DNA sequence-based scores to better understand not only if a genomic variant alters a protein, but how. We believe this is a critical step for understanding mechanistic changes due to genomic variants, designing functional validation tests, and for improving disease classifications. We applied this approach to the RAS gene family encoding seven distinct proteins and their 935 unique missense variants present somatically in cancer, in rare diseases (termed RASopathies), and in the currently healthy adult population. This knowledge shows that protein structure-based scores are distinct from information available from genomic annotation, that they are useful for interpreting genomic variants, and they should be taken into consideration in future guidelines for genomic data interpretation.\n\nSignificance StatementGenetic information from patients is a powerful data type for understanding individual differences in disease risk and treatment, but most of the genetic variation we observe has no mechanistic interpretation. This lack of interpretation limits the use of genomics data in clinical care. Standard methods for genomics data interpretation take advantage of annotations available for the human reference genome, but they do not consider the 3D protein molecule. We believe that changes to the 3D molecule must be considered, to augment current practice and lead to more precise interpretation. In this work, we present our initial process for systematic multi-level molecular scores, including 3D, to interrogate 935 RAS-family variants that are relevant in both cancer and rare diseases.

genetics↗

bioRxiv2019

Discovery, Expression, Cellular Localization, and Molecular Properties of a Novel, Alternative Spliced HP1g Isoform, Lacking the Chromoshadow Domain

By reading the H3K9Me3 mark through their N-terminal chromodomain (CD), HP1 proteins play a significant role in cancer-associated processes, including cell proliferation, differentiation, chromosomal stability, and DNA repair. Here, we used a combination of bioinformatics-based methodologies, as well as experimentally-derived datasets, that reveal the existence of a novel short HP1{gamma} (CBX3) isoform, named here sHP1{gamma}, generated by alternative splicing of the CBX3 locus. The sHP1{gamma} mRNA encodes a protein composed of 101 residues and lacks the C-terminal chromoshadow domain (CSD) that is required for dimerization and heterodimerization in the previously described 183 a. a HP1{gamma} protein. Fold recognition, order-to-disorder calculations, threading, homology-based molecular modeling, docking, and molecular dynamic simulations show that the sHP1{gamma} is comprised of a CD flanked by intrinsically disordered regions (IDRs) with an IDR-CD-IDR domain organization and likely retains the ability to bind to the H3K9Me3. Both qPCR analyses and mRNA-seq data derived from large-scale studies confirmed that sHP1{gamma} mRNA is expressed in the majority of human tissues at approximately constant ratios with the chromoshadow domain containing isoform. However, sHP1{gamma} mRNA levels appear to be dysregulated in different cancer types. Thus, our data supports the notion that, due to the existence of functionally different isoforms, the regulation of HP1{gamma}-mediated functions is more complex than previously anticipated.

molecular biology↗