wiener_deconv: Wiener deconvolution methods for estimating concentration dynamics from fluorescence traces

Noncausal Wiener deconvolution (deconv.py)

These are functions for computing Wiener deconvolution and denoised reconstruction of a 1D (e.g. fluorescence) time-series as in ref. 1. Here, concentrations are estimated using a time-domain representation of the impulse response for the fluorophore. The noise spectrum is assumed to be flat.

This code also includes a method for estimating the regularization parameter W based on desired absolute error.

Test notebook (test_deconv.py)

Notebook for testing the noncausal deconvolution method.

Fluorescence data (fluorescence_detrended.csv)

Detrended fluorescence data for use in the test notebook.

Causal Wiener deconvolution (causal_deconv.py)

Causal Wiener deconvolution was also implemented following the derivation in ref. 2. These methods were not used in ref. 1 but are included here for research purposes.

References

1. Neyhart, E et al. (2024; accepted) Cortical acetylcholine dynamics are predicted by cholinergic axon activity and behavior state. Cell Reports.
2. Lindell, Elisabeth. (2004) "On the Influence of Sensor Dynamics in Estimation of Reactor Noise Signals."