Precise calcium-to-spike inference using biophysical generative models [Biophys_SMC and Biophys_ML]
Year: 2026
Type: software
Abstract: # C-SPIKES Capsule ## This capsule contains the core elements of the Biophys_SMC and Biophys_ML methods. ### Component use is demonstrated in the code_ocean_inference_demo.py script, accessible through code/run.sh, which - Runs inference on a subset of the full benchmark dataset (a larger portion can be processed by passing the `--dataset-percent` flag with values from `0` to `100` where `100` indicates running all files) with the $Biophys_{SMC}$ (aka, the particle Gibbs with Ancestor Sampling or _PGAS_), $Biophys_{ML}$, $ENS^2$, and CASCADE methods - Outputs cached inference runs to `results/inference_cache for each method run through the default `./run.sh` stage with stored metadata, particle Gibbs outputs, and plots of pre-generated reference data directed to `results/inference_parity` - Produces summary comparisons of correlations of reference and generated runs with ground truth data in `results/inference_parity/correlations_parity.csv` ### More information about run options, outputs, as well as local build instructions can be found in the capsule root `README.md`, `kokkos_install.md`, and `README_CodeOcean.md` markdown files.
Source: Code Ocean
Institutions: Princeton University, Institut Pasteur
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7224136295