Projects
This page details some of the projects I worked on during my graduate education. Due to the nature of employment, projects I have worked on since leaving grad school are the property of the companies which funded the work. However, assuming I update my blog on a reasonably consistent basis (wishful thinking) then you'll also get a sense of what I have done since then. Of course, if you have specific questions about a line item on my CV, a particular skill, or a consulting opportunity, then please feel free to reach out (email preferred!).
Enabling Reproducible Bioinformatics Workflows
LoadExp+
As the prices for sequencing techologies have decreasaed many researchers are now able to conduct large experiments on the genomic and even metagenomic scales. This progression is exciting, and the need for computational expertise in the biological sciences has never been greater! However, the need for this experience has greatly surpassed the number of trained individuals, and not every lab wants to invest the time into learning these new technologies. They want to answer scientific questions!
I worked with Dr. Eric Lyons' lab at The University of Arizona to add analysis workflows to their web-based comparative genomics resource, CoGe. These workflows, collectively called "LoadExp+" allow researchers upload raw data from their local machine or from cloud storage by integration with the CyVerse Datastore. Output files are automatically organized into digital "notebooks" to be shared with colaborators or kept private if desired. Each experiment's files are also viewable as tracks in CoGe's integrated genome browser (Epic-Coge). This enables researchers to run LoadExp+ workflows using any available genome in CoGe, including those uploaded by users themselves.
Automated Analysis Pipelines With Snakemake and Singularity
While the simple solution of writing a BASH script for each step of a workflow does work, as you run more types of analyses this doesn't scale well. Your directory structure becomes a mess and you fail to make efficient use of available resources for a workflow (i.e. step 1 needs 8 cores, step 2 only needs 1, how do I leverage my 28 core system for this?). Even more concerning, when there are different versions of piecemeal shell scripts floating around and nobody knows which was used to for sample X.
Snakemake is simple and full featured workflow manager based on Python. It intelligently replaces processes in a workflow to paralellize jobs up to your maximum resources at run-time. Once a pipeline has been tested it can be packaged into a container with Singularity. Similar to Docker, Singularity is optimized for HPC environments. This keeps HPC sysadmins happy because they don't need to manage your software, plus you can port the analysis to any system you want!
Work is ongoing to implement our lab's pipelines for small RNA and whole-genome bisulfite sequencing in such a framework.
The Role of RNA-directed DNA Methylation in Seed Development
Maternal components of RNA‐directed DNA methylation are required for seed development in Brassica rapa
It's been widely observed that RdDM and 24 nucleotide small RNAs are correlated with reproductive processes. This is due to specific methylation patterns in reproductive tissues and the high abundance of small RNAs in seeds and flowers. Most of this work has been done in Arabidopsis thaliana a premier model organism in plant biology. However, A. thaliana's small, TE-poor, genome is not typical of most plants and therefore we are investigating the connection between plant reproduction and RdDM in Brassica rapa a closely related, recently outbreeding species which shows severe defects in seed development in RdDM mutants.
We have identified mutations in several RdDM pathway genes, annotated RNA Pol IV/RDR2-dependent small RNA loci, and determined which genes are differentially expressed in RdDM mutant ovules and seeds. We have also found that it is the maternal sporophytic genome that determines whether the seed developmental defect will be present.
You can check out the publication here!.
Abundant expression of maternal siRNAs is a conserved feature of seed development
Our finding that seed development in B. rapa depends on the RdDM machinery opens up many new avenues of discovery. We expanded upon this research by investigating the dynamics of small RNA expression and accumulation genome-wide in seeds. From this large collection of sequencing data we identified a subset of small RNA loci that account for the vast majority of small RNA expression in seeds. These loci were identified as siren loci, after a similar category that had been observed in rice endosperm.
Siren loci are unique amongst small RNA loci. They accumulate Pol IV-dependent small RNAs and direct DNA methylation, but in a tissue-specific manner. Additionally, siren RNAs tend to be expressed only from maternal chromosomes even when functional paternal RdDM machinery is present in the endosperm. Finally, siren loci and siren RNAs are dependent only on maternal RdDM machinery in the seed and are present in multiple species. This indicates that siren loci may be a general feature of seeds, and may be required for seed production in some species.