suntag-analysis

Extraction and analysis of SunTag traces, starting from time-lapse microscopy data and spatial trace annotations (obtained with TrackMate).

1) SpotFitting

Bayesian inference of the spot intensity for each trace.

Input

• time-lapse image

• positions of the spot in each trace (TrackMate output file)

Output

• csv file for each trace and channel, containing the intensity of each spot.

To run on a server in a distributed way (multiple acquisitions):

bash 01_submit_job.sh

2) SaveDataAsJSON

Merging files containing the trace intensities into one unique JSON file that will be the input of Bayesian inference.

3) EstimateNoise

Estimating the noise from untranslated traces. This parameter will be used as input in the HMM.

4) HMM

Bayesian inference with Hidden Markov Model (HMM)

Input

• JSON file contaning the GFP intensity traces for control and perturbed conditions
• Stan model (.stan)

Output

.pkl file containing the inferred model parameters

Two different models can be used

• RB2_Sigma_time.stan: models where the mature protein intensity (u) is a model parameter
• RB2_Sigma_time_uFixed.stan : model where the mature protein intensity is a fixed parameter, given as input