We create software packages for you that contain scripts/notebooks/plugins that are native to your HPC/Desktop/Laptop computing environments. We specialize in the following:
1) Create custom training data for 3D segmentation of nuclei, membranes, and specialized structures in multiple channels.
2) Train custom 3D deep learning models using HPC resources, including StarDist, CARE, U-Net, VollSeg, AutoEncoder, CellPose, and similar models.
3) Create custom training data and ONEAT models to detect cell states such as mitosis, apoptosis, necrosis etc in 3D time-lapse datasets.
4) Use biological knowledge about mitosis and apoptosis locations over space and time to automatically correct lineage trees in TrackMate.
5) Convert TrackMate XML files into DataFrame objects with shape and dynamic information from point clouds or autoencoder models, with optional latent space features.
6) Build unsupervised, XGBoost, or CNN-attention models to classify cell fate based on morphodynamic information.
7) Customize Inception-style models to improve cell fate prediction accuracy, even for classes with few examples.
8*) Apply concepts from non-equilibrium statistical mechanics to single-cell trajectories to gain physics-based insights into developing systems.
9) Develop custom analyses on request, such as recoil velocity calculation, mean squared displacement analysis, or creating custom Napari plugins.
* Only for those paid collaborators who have done all the previous steps with our provided software package.
