Deepcell has recently announced to release of three large cell morphology data sets to permit scientists to examine unique high-dimensional morphology data.
The data packages will be released at the Advances in Genome Biology and Technology (AGBT) General Meeting which will be held in Hollywood, Florida. Here the Deepcell will make all the new data sets available for offline and online usage.
Deepcell is a leader in AI-powered single-cell research that helps drive fundamental biological breakthroughs. These data sets were produced using the high-throughput platform from Deepcell, which consists of imaging and sorting equipment, AI modeling, and algorithms.
To detect a tumor, immunological, and epithelial cell types in the first data set, the Deepcell platform was applied to a variety of human melanoma cell lines and primary tumor samples, relying solely on morphology.
To obtain more clarity into this morphologically unique subpopulation, the melanoma tumor cell population statistics from this time series were then chosen in the Deepcell software package and re-projected using a bespoke UMAP to generate a second data set.
This makes visible the heterogeneity among these cells based on minute morphological variations, such as pigmentation, which might be challenging to spot using traditional techniques.
In the final data set, a range of human dissociated tumor cell (DTC) datasets was used to examine the morphological heterogeneity of immune cell types in the lung tumor microenvironment using Deepcell label-free methodology.
The Human Foundation Model (HFM) AI model has been trained using millions of cell images of various stages and types. Scientists may readily generate high-dimensional readouts of well-known and novel morphological traits from unlabeled cells using an unbounded hypothesis method thanks to this model.
To classify live cells for subsequent molecular or functional research, the software suite also enables the generation of customized cell classifications and the detection of visually comparable cell groups.