High Resolution Comparison of Cancer-Related Developmental Processes Using Trajectory Alignment

Abstract

Abnormal differentiation is a key feature of cancer, yet currently there is no framework that enables a comparative analysis of differentiation processes across patients while preserving their individual-level resolution. Here, we present devMap, an algorithm that uses high-dimensional trajectory alignment to anchor cancer-related developmental processes to a common backbone process, thus allowing for their systematic comparison. We applied devMap to bone marrow samples from healthy individuals and AML patients profiled by single-cell mass-cytometry at cancer diagnosis and following treatment. devMap standardization enabled us to infer the developmental status of the AML samples and characterize its evolution following treatment and in relapse. Application of devMap on an external dataset of AML bone marrow samples revealed conserved patterns of developmental signaling responses in AML that were obscured by traditional methodologies for developmental inference.