Abstract

Single Cell Phenotype Analysis and Isolation based on Image-Guided Cell Sorting

Yu-Hwa Lo, Professor, University of California San Diego

We present unique technologies that bridge single cell genomics and single cell phenotype analysis.  Tissue samples can contain millions of cells.  Even with high throughout sequencing, the NGS technologies and analysis tools may handle 1000 to 10,000 single cells for each sample due to cost and throughput considerations.  This corresponds to a very low (~0.1%) sampling ratio and increases the risk of missing important rare cells in the sample.  Currently people use cell sorting according to biomarkers to avoid the under sampling problem.  However, recent studies have shown that there exist far more cell types and degrees of heterogeneity, especially in tumorous tissues, than we have assumed before. The limited number of biomarkers and specificity of these markers have compromised our abilities to classify and sample cells in order to produce complete, unbiased understanding of the most interesting and significant cells in the population.  

To address this bottleneck, we developed high throughput flow cytometer cell sorting systems that can analyze and isolate single cells based on their phenotypic characteristics such as cell volume, shape, morphology, nuclear features, organelle distribution, speckle counts, etc.  The systems can detect both fluorescent and label-free (scattering) geometric features.  Coupled with machine learning and neural network, the systems offer capabilities to select and isolate single cells of different cell types and cell states, and to normalize the population (i.e. to select equal number of cells of the same type) to remove bias in the downstream genetic analysis.