Deepcell Releases Sample Datasets to Highlight Morphology-Driven Data
By Bio-IT World Staff
February 7, 2023 | Deepcell, a pioneer in AI-powered single cell analysis to fuel deep biological discoveries, has released three datasets, available online, to enable researchers to explore novel high-dimensional morphology data. The datasets were generated on Deepcell’s high-throughput platform, comprised of imaging and sorting instrumentation, AI models, and a software suite.
The AI model, Human Foundation Model (HFM), has been trained on millions of cell images and enables scientists to easily produce high-dimensional readouts of known and novel morphology features from unlabeled cells in an unbounded hypothesis approach. The software suite—which will also be available in this data release—allows for the creation of custom cell classifications and identification of morphologically similar cell groups for sorting of viable cells to enable downstream molecular or functional analysis.
This first set of data releases showcases how the Deepcell technology can characterize different cell types in a heterogeneous sample in a label-free manner and allows the user to analyze specific cellular populations of interest that are difficult to identify with molecular markers.
Close Look at Morphology
Three human cancer datasets are available for exploration, all comprised of samples purchased from well-known vendors such as Discovery Life Sciences, Cell Biologics, and BioIVT, Giovanna Prout, VP of Marketing, explained in an email to Bio-IT World.
In the first dataset, the Deepcell platform was used on 28 samples, a mixture of human melanoma cell lines and primary tumor samples to identify tumor, immune, and stromal cell populations in a label-free manner, using only morphology. A set of about 28,000 cells are used in the UMAP projection through the Deepcell Software Suite for ease of visualization purposes, Prout said.
The melanoma tumor cell population data from this dataset was then selected in the Deepcell software suite and re-projected using a custom UMAP in order to gain additional resolution into this morphologically distinct subpopulation to create a second dataset of about 350,000 cells. A set of about 18,500 cells are used in the UMAP projection through the Deepcell Software Suite for ease of visualization purposes, said Prout. “This reveals the heterogeneity within these cells based on subtle morphological distinctions, including pigmentation, which can be difficult to identify using conventional methods,” she noted.
In the final data set, the Deepcell label-free technology was used to explore the morphological diversity of immune cell populations in the lung tumor microenvironment from a variety of human dissociated tumor cell (DTC) samples. The third dataset is a compilation of 11 samples run on the Deepcell platform for a total of about 1.1 million cells. A set of about 17,000 cells are used in the UMAP projection through the Deepcell Software Suite for ease of visualization purposes, Prout said.
“Morphological variations, even within one sample, are staggeringly high and nuanced. No predetermined list of features or classes are descriptive enough to capture this wealth of information. We believe that our approach to foundation models and self-supervised learning provide an unparalleled advantage in studying cell morphology," said Mahyar Salek, President, CTO and Founder of Deepcell in a press release announcing the availability.
Deepcell hopes the data release will help the research community visualize the high dimensional single cell morphology data generated on Deepcell’s platform. “We expect researchers to explore the datasets to understand the type of analyte Deepcell generates through its platform and peak their interest in where this data can be used for in the context of examples/ biological scenarios,” Prout said. She hopes that the datasets will prompt new ideas about how the technology could be used to tackle new research.