RESEARCH INTERESTS
Single-cell multi-tissue RNAseq and ML modeling
Fig. 4 from Poster Presented at AGT Conference 2024
A) UMAP comparison of HV to PC in the PBMC T-cell. B) UMAP comparison of HV to PC for T-Cells in CSF. C) Bar plot comparing the proportions for cell types in HV to PC in the PBMC and the proportion of cells per sample clustered by similarity. D) Bar plot comparing the proportions for cell types in HV to PC in the CSF and the proportion of cells per sample clustered by similarity. E) Bar plot with the proportion of clones broken down by sample and class (HV vs. PC) for PBMC. F) Bar plot with the proportion of clones broken down by sample and class (HV vs. PC) for CSF. G) Comparison of shared clones between compartments(CSF and PBMC) for two samples. PC = neuro-PASC , HV= Healthy Volunteers
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes Long-Haul COVID, which affects 1 in 4 infected individuals worldwide. Of these, 10% experience ongoing symptoms lasting months to years, including neurologic issues such as brain fog, anxiety, depression, sleep problems, and headaches, even after viral RNA becomes undetectable. The molecular and immunologic mechanisms driving these persistent neurologic symptoms remain unknown. To address this, we characterized the transcriptome of peripheral blood mononuclear cells (PBMCs) and paired cells isolated from cerebrospinal fluid (CSF) of individuals with neurologic post-acute sequelae of SARS-CoV-2 infection (neuro-PASC) using 10X Genomics single-cell RNA (scRNA-seq) sequencing. I implemented a standard scRNA-seq analysis pipeline using Seurat and CellChat in R to identify gene expression differences and intercellular communication patterns. I then developed a custom machine learning pipeline applying Partial Least Squares (PLS) modeling to determine whether the model could distinguish healthy individuals from long-haul COVID patients, and to identify which genes within each cell type were most influential in differentiating the two populations.
Phylogenomics of Bonytongues (Osteoglossomorpha)
Fig 1. from Peterson et. al, 2022: (a-d): previous morphological and molecular phylogenetic hypotheses for Osteoglossomorpha. Inter-continental sister-group relationships implied by these phylogenies include (e): Maximum Likelihood tree for Osteoglossomorpha based on a concatenated 546 exon dataset (this study). (f): Pantodon buchholzi (Pantodontidae); (g): Osteoglossum bicirrhosum (Osteoglossidae); (h): Chitala chitala (Notopteridae); (i): Gymnarchus niloticus (Gymnarchidae); (j): Mormyrops boulengeri (tubesnout); (k): Gnathonemus petersii (Schnauzenorgan); (l): Marcusenius stanleyanus (chin swelling; CUMV 96481); and (m): Campylomormyrus mirus (tube snout with Schnauzenorgan).
Convergence of Cranial Facial in African Weakly Electric Fish (Mormyridae)
Osteoglossomorpha (Bonytongues) represent an enigmatic lineage of early-diverging teleost fish. This order contains clades such as the African Weakly Electric Fish (Mormyridae), the Featherbacks (Notopteridae), the African Butterflyfish (Pantodontidae), and Arapaima (Osteoglossidae). Our study combines genomic markers with six fossil calibrations providing unprecedented resolution for this clade. Noteworthy, our phylogeny resolves the position of the African Butterly fish (Pantodontidae), typically a rogue taxon, as sister to all osteoglossiforms. Our study provides a phylogenetic framework to investigate the evolutionary history of Bonytongues.
African Weakly Electric Fish (Mormyridae) are characterized by five unique character states chin swellings, schnauzenorgan, normal, tubesnouts, and tubesnouts with schauzenorgan. These character states are hypothesized to either enhance foraging behavior or amplify electrosensory abilities. These different appendages have evolved multiple times independently within African Weakly Electric Fishes. Therefore, to understand how and why these traits might be convergent, these character states will be investigated with both histology and comparative genomics.