|Ph.D Student||Alpert Ayelet|
|Subject||Characterization of Dynamical Processes through|
High-Dimensional Profiling Yields Meaningful
|Department||Department of Medicine||Supervisor||Professor Shai Shen-Orr|
|Full Thesis text|
Biological systems constantly sense and respond to environmental stimuli by changing biologically relevant sets of features. Recent computational advances allow a high-resolution characterization of such dynamical processes using high-dimensional data collected on single objects exhibiting variable developmental states through their ordering along developmental trajectories.
In the first part of this work, we used multiple ‘omics’ technologies to capture population- and individual- level changes in the human immune system of 135 healthy adult individuals of different ages sampled longitudinally over a nine-year period. We observe a high inter-individual variability in the rates of change of cellular frequencies that correlate with baseline values, allowing identification of steady state levels towards which a cell subset converges and the ordered convergence of multiple cell subsets towards an older adult homeostasis. These form a high dimensional trajectory of immune-aging (IMM-AGE) that describes a person’s immune status better than chronological age. We show the IMM-AGE score is associated with cardiovascular disease and predicts all-cause mortality beyond well-established risk factors in the Framingham Heart Study.
In the second part of this work, we developed the cellAlign algorithm that aligns two different processes characterized at the high dimension by single-cell trajectories. cellAlign global alignment aligns the whole trajectories thereby provides the overall similarity between processes whereas cellAlign local alignment identifies regions along the two processes exhibiting similar features’ expression dynamics. We validated the cellAlign algorithm using trajectories of single-cell mass cytometry and RNA sequencing data and showed it is robust to the noise typically observed in these data. We further leveraged the cellAlign algorithm to identify differences in gene expression dynamics along human and mouse embryogenesis, and to derive the sequential cascade of genes upregulation along the two processes.
In the last part of this work, we developed tuMap, a computational algorithm that exploits high-dimensional single-cell data of cancer samples exhibiting an underlying developmental structure to align them with the healthy development yielding the tuMap pseudotime axis that allows their systematic comparison. We applied tuMap on single-cell mass cytometry data of acute myeloid leukemia (AML) patients to quantify changes in cellular abundances at the time of AML diagnosis, following treatment, and in relapse. We further utilized tuMap to study signaling alterations in AML using external datasets of single-cell mass cytometry and gene expression. By aligning cancer single cell data, tuMap facilitates data integration and yields a high-resolution metric for cancer classification.
Collectively, our work utilizes the detailed characterization of biological processes obtained by single-cell trajectories to define different kinds of quantitative frameworks. Specifically, we developed a metric quantifying the biological age of the immune system and a metric of processes that was further validated on cancer samples. As quantitative comparison is a fundamental analytical tool and the increased use of high-dimensional profiling to characterize biological processes, we believe these methodologies are valuable tools for large-scale biological systems research.