flowLearn

flowLearn automates semi-supervised identification and quality checking of cell populations in flow cytometry to enable reproducible gating across samples.

Key Features:

• Semi-Supervised Learning: Employs semi-supervised learning using a small set of manually gated samples to generalize gating across new datasets.
• Density Alignments: Uses density alignments to align cell distribution densities between training and new samples for gate prediction.
• Minimal Manual Training: Predicts gates across samples using only a minimal number of manually gated training samples.
• High Accuracy: Achieves median F1-scores exceeding 0.99 for 31% and 0.90 for 80% of analyzed populations across two state-of-the-art datasets.
• Interpretability and Adjustability: Generates interpretable automated gates that are adjustable based on initial training data.

Scientific Applications:

• Immunology: Automates and standardizes gating for immunophenotyping and immune cell population analysis in flow cytometry studies.
• Oncology: Supports cancer and tumor immunology studies requiring accurate cell population identification and quality checking.
• Stem Cell Research: Facilitates quantification and quality control of stem cell populations in flow cytometry experiments.

Methodology:

Semi-supervised learning using density alignments to align cell distribution densities between manually gated training samples and new samples to predict gating boundaries.