February 13, 2024
3:00 pm – Refreshments
3:30 pm – Presentation
Speaker:
(1) Ted Kalbfleisch, Department of Veterinary Sciences, University of Kentucky
(2) Christopher Crawford, Department of Physics and Astronomy, University of Kentucky
Where:
Davis Marksbury Building – James F. Hardymon Theatre
(Zoom link: https://uky.zoom.us/s/84474671604)
Title:
(1) The Challenges and Opportunities of Reference Free Genomic Data Analysis
(2) Real-time least-squares pulse fitting for digital spectroscopy
Abstract:
(1) Since its inception, the cost of generating high quality genomic sequence has plummeted toward zero. Pipelines that perform primary analyses and genome assembly have scaled well with available computing power and have rarely been a rate limiting step for genomic analyses. Historically, a species has had a single reference genome which has been well annotated for gene structure and epigenetic marks. Sequence data for any animals from that or a closely related species will be mapped to that reference for analysis in that curated genomic context. The advent of inexpensive long, and ultralong sequencing technologies have made the assembly of phased, nearly gapless genomes routine. The main advantages are that we can study genomes fully phased as they are in nature, and large structural variants in ways that have been impossible previously. This has created a new paradigm where each of these new genomes will need to be analyzed in a context not of a reference genome, but rather in the larger context of all that is known. The co-emergence of this challenge and artificial intelligence creates a tremendous opportunity. With the insights of computer science colleagues, we can now begin to ponder the data structures that will be most amenable to analysis by AI technologies, as well as the metadata that include both molecular, and macroscopic phenotype data, and how these datasets could be readily accessed and analyzed in a distributed network of both computing and storage.
(2) With the availability of cost-effective digitizers and powerful pipeline processors, modern spectroscopy can be performed completely in the digital domain with minimal front-end analog signal processing. These systems offer the flexibility and extensibility of digital signal processing algorithms to simultaneously extract multiple waveform parameters such as pulse height and start time. However, up until now, the computationally intensive task of pulse fitting has traditionally been performed offline on computing clusters to obtain the final spectra. I present new algorithms developed to perform these least-squares fits to template waveforms in real time on Field Programmable Gate Arrays (FPGAs) and GPUs. They will be used to enhance online spectroscopy of gamma rays in nuclear experiments with up to 10^16 events to analyze.
Slides:
(1) Ted Kalbfleisch – CCS_Kalbfleisch_2024_02_13.pdf
(2) Christopher Crawford – sls_conv_2024-02-13_ccs.pdf