EfficientNet-Based Embeddings of Single-Cell E. coli Gene Deletion Strains from Multichannel Microscopy

Dataset Description This dataset contains single-cell embeddings of Escherichia coli (E. coli) derived from multichannel microscopy images using a deep learning model (EfficientNet). The model was trained to perform binary classification between wild-type and mutant (gene deletion) cells. The dataset includes embeddings for over 800,000 individual cells, covering 107 gene deletion strains and 163 wild-type replicates. Each cell is represented by a feature vector of 1,280 values extracted from the last convolution layer of the EfficientNet model, plus the cell pixel area. The microscopy images used to generate these embeddings include four imaging channels: C1: Phase Contrast C2: Nucleoid (DAPI) C3: FtsZ (Venus) C4: SeqA (mCherry) File Format and Extraction The dataset is provided as a compressed .tar.gz archive. To extract the contents, run the following command in a Unix-based terminal: tar -xvzf your_dataset_name.tar.gz