Measuring Reproducibility of High Throughput Experiments and LLM Generated Analyses

Explore methods for measuring reproducibility in high-throughput experiments and LLM-generated analyses through this 42-minute conference talk from the Computational Genomics Summer Institute. Learn about statistical frameworks for assessing reproducibility across various experimental contexts, including approaches for handling missing data and covariate effects. Discover the HiCRep method for evaluating Hi-C data reproducibility using stratum-adjusted correlation coefficients, and examine regression frameworks that account for experimental variables affecting reproducibility. Understand the challenges posed by irreproducible research findings in scientific literature and gain insights into novel analyst-inspector frameworks for evaluating the reproducibility of large language models in data science applications. Access supporting research papers covering foundational concepts in reproducibility measurement, specialized methods for genomic data analysis, and cutting-edge approaches to LLM evaluation in computational biology contexts.