Maryam May

This paper introduces a neural network-based batch alignment method for cytometry data that preserves biological signals and enables hyperdimensional representation for improved disease phenotype classification.

The paper benchmarks pairwise distance metrics for classifying mouse mutant phenotypes from flow cytometry data, finding cosine similarity and standardized Euclidean distance perform best.

The paper quantifies how functionally constrained human proteins are less affected by stability-disrupting missense mutations and shows their depletion of extreme stability-changing variants in disease-free populations.