Bayesian nonparametric lifetime analysis (BNP-LA) is a software packaged developed base on the framework described 
in "Fluorescence Lifetime: Beating the IRF and interpulse window" by Mohamadreza Fazel, Alexander Vallmitjana, 
Lorenzo Scipioni, Enrico Gratton, Michelle A. Digman and Steve Presse, Biophysical Journal (2023). 

BNP-LA is capable of dealing with a broad range of lifetimes and photon ratios using as few photons as 500 acquired 
using a pulsed illumination procedure. The MATLAB software package comes with a gui and here we illustrate how to 
work with the gui. To run the algorithm you need to have matlab 2020 or higher versions installed with the 
Statistics and Machine Learning Toolbox available.

To open the gui set your MATLAB path to the Functions-folder, type "gui" in the command window and hit "Enter". 
The gui has 4 sections:

1) Exp Params: this section contains calibrated parameters from the experimental setup:
          Interpulse time: the period between two consecutive laser pulses in ns (Default: 12.8ns)
          IRF mean:        mean of the Gaussian IRF in ns (Default: 12.2ns)
          IRF sigma:       sigma of the Gaussian IRF in ns (Default: 0.66ns)
          
2) Method Params: parameters of the analysis method which are mostly good as they are
	  Max Species:     maximum number of species (described by M in the paper) to be considered in the
	  		   Dirichlet process. This should be larger than the number of species present in 
	  		   the input data (Default: 5)
                           N.B. Setting too high can increase computational cost.
	  Max Pulses:      maximum number of pulses following the excitation pulse over which a fluorophore might 
		           remain excited (Default: 5) 
                           N.B. Setting too high can increase computational cost.
	  Alpha:	   parameter of the gamma prior on lifetimes (Default: 1)
	  Beta:	           parameter of the gamma prior on lifetimes (Default: 50)
	  
3) Section with no title: 
	  Photon Count:    number of photons to be used within the analysis (must be smaller than the total
                           input photons) (Default: 5000)
	  Iterations:      number of samples to be taken from the posterior (more samples tend to give better 
                           results but adds computational cost) (Default: 2500)
	  LOAD DATA:       this button popups a directory where data file can be selected. The data should be an  
                           array of photon arrival times in "ns" and saved as a row vector called "Dt".
	  RUN:             this button runs the algorithm
	  
4) Results:
	  Photon Ratio Threshold: this is a key parameter in plotting results. All plots will show histograms
                                  of samples with photon ratios (weights) larger than this threshold value
	  Parameter:              pick one of the options from the popup menu to be shown. There are four options:
			          a) "plot ratio" will show a plot of all the photon ratio samples vs Monte Carlo 
                                     iterations (good at assessing convergence) taken from the posterior.
                                  b) "plot lifetime" same as above but for the lifetimes.
                                  c) "hist lifetime" will show the histogram of the sampled lifetimes associated to
                                     the species with photon ratios larger than the threshold given.
                                  d) "hist ratios" the same as item c but for photon ratios.
	  PLOT:                   this button will plot the results for the parameter picked above
	  
After the analysis is done, the time that it took to analyze the data will be printed in the command window. We 
would suggest to first look at the plot of the photon ratios to detemine the number of species. This number is given 
as the number of species with non-negligible photon ratios in the resulting plot. Next, use the photon ratio plot to 
set a threshold value and plot the lifetime histograms to see the found lifetimes. 

The package contains a set of simulated data with lifetimes of 0.2ns and 0.6ns (both below the default IRF sigma) 
with the default parameters in the gui. Use the load button to load the data and hit RUN. It will take less than 30s 
to analyze the data. Next, select one of the parameters from the popup menu in the "Results" section and hit PLOT.   
Note: the default number of iterations is 2500 but this value is data dependent and some data sets might need larger 
number of iterations. Specifically, data sets containing 3 or more species or more similar lifetimes require more 
iterations (samples) which, in turn, leads to more computational complexity. The other data set is in vivo 
experimental data containing three species. For a description of this data see the following section.

The script "GenData.m" can be used to simulate different data sets. The existing parameters will generate the 
simulated data given in the Data-folder.

The function "findIRF" can be used to calibrate the parameters of a Gaussian IRF, namely mean and standard 
deviation, using a data set containing a single species with known lifetime.

If you happen to have any problem please shoot us an email at fazel.mohamadreza@gmail.com or spresse@asu.edu
	   
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Experimental Data Description:
Experimental data were acquired on an ISS-Alba confocal microscope. Excitation was provided by a 780 nm 2-photon laser pulsed at 80 MHz (Calmar Laser). Emission was collected by an avalanche photodiode (Excelitas Technologies) and an ISS A320 Fast FLIM unit for lifetime determination. A 760 short-pass dichroic was used to split the excitation and emission light; no additional emission filter was used. Images were collected with a 16 μs pixel dwell time with an image size of 256 × 256 pixels and a 0.39 μm pixel size, accumulating 7–9 frames for each data set. After each experiment, the images were preprocessed by extracting individual photons and constructing a list of arrival times for each photon, both in reference to the start of the measurement and to the previous laser pulse. MDA-MB-231 cells were seeded in a glass-bottom 8-well plate (Ibidi GmbH) previously coated with 2 μg/mL Fibronectin in Dulbecco’s phosphate buffer solution (DPBS) without Ca, Mg, Phenol Red (GenClone, Genesee). Cells were stained with 50 μg/mL pHrodo Green Dextran, 100 nM TMRM, and 50 nM LysoTracker Deep Red (Invitrogen) for 4 h and subsequently imaged; see Fazel, et al, ACS Photonics 2023. 
For this data set, the parameters are slightly different from the default values in the gui and given as: Interpulse times = 12.5ns, IRF mean = 12.22ns, IRF sigma = 0.8ns. To analyze this data set we suggest using 20K-30K photons, 5K-10K iterations and a threshold of 0.1. The lifetimes for these fluorophore species are ~0.6ns, ~2.4ns, and ~4.5ns.