Driven by recent advances in single-cell sequencing, our understanding of the cellular profiles in healthy and pathological tissue is rapidly expanding. However, understanding development and disease will require linking molecular cell types to their functional correlates. Our lab is focused on developing novel high-throughput technologies that can perform “functional” measurements at single-cell resolution, and link them to the molecular profile (e.g. transcriptome, phenotype) of the cell. The single-cell platforms that have been developed in our lab include TRAPS-seq (Nature Methods, 2023) for linking single-cell cytokine secretion with transcriptome, PAINTKiller-seq (Nature Biomedical Engineering, 2024) for joint profiling of single-cell cytotoxicity with transcriptome, Drop-PEARL (Lab Chip, 2022) for screening of cells secreting virus-neutralizing antibodies, and TRuST (Advanced Science, 2023) for label-free selection of functional cells for immunotherapy. Ultimately, by integrating these multidimensional measurements, we hope to model the single cell as a system of interacting networks at multiple scales. These cellular models, as the basis of “Single Cell Systems Biology”, will be instrumental in our understanding and prediction of how a cell changes over time and under varying condition. This will create potential for entirely new kinds of explorations, including drug designs to improve cellular function and eliminate diseases.

Precision medicine relies on highly-multiplexed molecular profiling tests for personalized diagnosis and patient stratification. However, the high cost and slow turnaround of next-generation sequencing hinder its practical implementation. Building on our core expertise in microfluidics and molecular biology, a research direction in my lab is developing rapid and affordable alternatives to next-generation sequencing for single-molecule measurements. My lab has spearheaded whole-new classes of single-molecule assays by developing the dqPCR (digital real-time PCR) and dmPCR (digital melting PCR) platforms. dqPCR allowed us to measure heterogeneity in single telomere lengths (Science Advances, 2020) as a prognostic marker in diseases such as pediatric neuroblastoma. On the other hand, multicolor dmPCR is a powerful concept that increases the multiplexing capability of digital PCR from 3-5 plex to > 100-plex. We demonstrated a wide variety of applications for dmPCR including microbiome profiling (Small Methods, 2022), pooled testing for infectious diseases (Small, 2023), and gene panel profiling (Advanced Science, 2023). These next-generation digital PCR platforms are promising technologies for enabling practical clinical implementation of precision medicine.

Epigenetics is the study of non-genetic DNA changes in response to the environment and plays an important role in modern diseases related to lifestyles. To enable the measurement of DNA methylation in limited amounts of samples, our lab developed various restriction-enzyme-based DNA methylation assays. We started with SCRAM (Science, 2013) and scGEM (Nature Methods, 2016), which are targeted DNA methylation assays on single cells. Subsequently, this evolved into the DARE assay (Nucleic Acids Research, 2019), which is capable of single-cell whole-genome DNA methylation profiling. More recently, we showed that the DARESOME assay (Science Advances, 2023) can be used for simultaneous measurement of 5mC and 5hmC from single cells. We are also interested in improving DNA methylation profiling for liquid biopsy. Drawing from our strengths in single-cell epigenetics, we tackled the challenges of detecting cancer through DNA methylation patterns in cell-free DNA from blood. Heatrich-BS (Science Advances, 2022) is a biomedical engineer’s solution to DNA methylation profiling in cell-free DNA. Recognizing that cancer markers are found in CpG-rich DNA regions, and the high GC-content bestowed these fragments additional thermal stability, we developed a simple heating protocol that efficiently enriched this subset of cell-free DNA. This strategy allowed us to profile each sample for ~US$30, an order of magnitude cheaper than conventional methods.