GUIDELINES for using the spatial dispersion analysis codes

Sebastien Herbert [email protected]

---

Matlab version R2017b

Required Matlab libraries

• https://fr.mathworks.com/matlabcentral/fileexchange/10867-uipickfiles-uigetfile-on-steroids

---

For cell type dispersion in a dynamic patern:

Preprocess the data files => extractDuplicateAndFormatDyn.m

Reformat the cell types exported in .csv format by Imaris into a single MATLAB file using the extractDuplicateAndFormatDyn.m function. The processing also checks that there are no duplicates in the files.

• This function will ask for the 3 expected cell types. Look at the csv files for an example. (Attention: there can be more than 1 input file for the type 1 population, mother cells are duplicated into the type2 population)
• Each file will be converted into a MATLAB table concatenated with the others and tested for duplicates. Duplicates are of 3 kinds:
  • duplicated cells are type2 and mother
  • duplicated cells are of the same cell type
  • duplicated cells are of different cell types (not mother) => only this situation will bring the function to throw an error message prompting the user to check which format should used (usually by going back to Imaris).
• If the files were properly set and provided, the output should be an 8 variable table by N lines (being the number of cells). Variables are :
  • PositionX
  • PositionY
  • PositionZ
  • Unit
  • time
  • ID
  • cellType
  • oldID (original Imaris cell ID, not always unique since the cells were concatenated).
• The table is saved under the name liveDataCurated.mat and contains the fullDataLive table

Launch the analysis of the dynamic pattern => pointDynPatternMain.m

This function will both handle the nearest neighbour analysis with simulation of the possible spatial distribution effects and their display. This function will automatically launch a parallel pool to accelerate calculations. The resulting analysis is stored under the completeAnalysis.mat file. By modifying the parameters in the code you can specify:

• which populations should be tested against which (For example mother against mother by permuting into the type 1+2 population).
• the number of random permutations for each model (10000 by default)
• the ranges of R and S to model
• that analyses were already calculated (if you just want to re-plot the displays), the algorithm will then ask you to select the analysis file (completeAnalysis.mat).

Once you are satisfied with the parameters to use, simply call the function and select the livedataCurated.mat created at the previous step. By default, this function will run output 4 images in .fig format and 4 data files in .mat format.

Images: Name is automatically appended with the permutation population

• RMSEisomap_permutIn_type1type2.fig: RMSE map with isolines separation
• RMSEmap_permutIn_type1type2.fig: RMSE map with interpolation between sampled positions
• allDistFreq_type1type2.fig: All distances frequency between the source and target populations
• averageDistFreq_type1type2.fig: All distances frequency between the source and target populations averaged per individual simulation

DataFiles:

• completeAnalysis.mat: Contains the next three files but can be quite big.
• NNanalysis.mat: Contains only the processed and analysed dispersion results
• NNdispersion.mat: Contains only the dispersion results prior to analysis
• NNdistances.mat: Contains only the distances

---

For cell type dispersion in a static patern:

• Entry point is pointPatternMain.m. All the relevant parameters are hard coded in the code.

• In order to analyse the same data from a dynamic case to a static case, you have to preprocess the livedataCurated.mat file into a new one, fit for the static analysis. To do so, use the formatDataDynToStat.m function

• call4NNDisplay.m: To display the RMSE map, save the map table in a simpler format and display the experimental vs CSR cdfs

• extract_data_bioCompare.m: To display a subplot figure of all the selected samples