A stochastic hybrid model of a biological filter

TL;DR

This paper introduces a hybrid stochastic model of a biological filter based on a genetic circuit, combining continuous protein levels and binary gene states, with an efficient inference algorithm demonstrated on simulated data.

Contribution

It presents a novel hybrid model of a genetic circuit with an associated efficient inference algorithm for partial observations.

Findings

Efficient approximate inference algorithm developed.

Model effectively captures biological filter dynamics.

Validated on simulated data.

Abstract

We present a hybrid model of a biological filter, a genetic circuit which removes fast fluctuations in the cell's internal representation of the extra cellular environment. The model takes the classic feed-forward loop (FFL) motif and represents it as a network of continuous protein concentrations and binary, unobserved gene promoter states. We address the problem of statistical inference and parameter learning for this class of models from partial, discrete time observations. We show that the hybrid representation leads to an efficient algorithm for approximate statistical inference in this circuit, and show its effectiveness on a simulated data set.