blinx

Introduction
Installation
Examples
Citation

Introduction

This repository contains code to estimate the number of fluorescent emitters when only their combined intensity can be measured, as reported in A Bayesian Method to Count the Number of Molecules within a Diffraction Limited Spot pre-print now available on BioRxiv. Experiments associated with the pre-print can be found in blinx_experiments

Detailed documentation can be found: here

Installation

For a basic CPU installation:

conda create -n blinx python
conda activate blinx
git clone https://github.com/funkelab/blinx.git
cd blinx
pip install .

For a GPU installation specific versions of jax and jaxlib must be pinned:

conda create -n blinx python cudatoolkit=11.4 cudatoolkit-dev=11.4 cudnn=8.2 -c conda-forge
conda activate blinx
git clone https://github.com/funkelab/blinx.git
cd blinx
pip install .
pip install 'jax==0.4.1' 'jaxlib==0.4.1+cuda11.cudnn82' -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html

Examples

blinx can be used for both inference and simulation. The estimate module contains functions to run inference on an intensity trace and determine the molecular count, while the generate_trace function can be used to simulate traces with known parameters.

Inference:

blinx.estimate

import blinx
from blinx.estimate import estimate_y

traces = ... # load traces

# specify the range of an initial parameter grid search
parameter_ranges = blinx.ParameterRanges()
# Specify hyper-parameters
hyper_params = blinx.HyperParameters()

count, map_parameters, likelihood, evidence = estimate_y(
    traces=traces,
    max_y=..., # maximum count to test
    parameter_ranges=parameter_ranges,
    hyper_parameters=hyper_params)

Simulation:

blinx.trace_model.generate_trace

import blinx
from blinx.trace_model import generate_trace

# Specify kinetic and intensity parameters
parameters = blinx.parameters.Parameters()
# Specify hyper-parameters
hyper_params = blinx.HyperParameters()

sim_trace, sim_zs = generate_trace(
    y=4, # the number of emitters
    parameters=parameters,
    num_frames=4000, # length of the simulated trace
    hyper_parameters=hyper_params)

Citation

If you use blinx in your research, please cite the BioRxiv pre-print:

@article{hillsley_bayesian_2024,
    title = {A Bayesian Solution to Count the Number of Molecules within a Diffraction Limited Spot},
    author = {Hillsley, Alexander and Stein, Johannes and Tillberg, Paul W. and Stern, David L. and Funke, Jan},
    doi = {10.1101/2024.04.18.590066},
    publisher = {{bioRxiv}},
    date = {2024-04-22},
}

funkelab / blinx