Image data analysis of synthetic unipolarity in E. coli
SynPolarityAnalysis is a Python script for analyzing spatial organizing protein distribution in bacterial cell.
- Overview
- System Requirements
- Installation Guide
- Data Acquisitions
- Data Input
- Execute Data Analysis
- Demo
- License
Our script serves as an interface containing self-defined scores and statistical tools to visualize the characteristics of spatial features such as polarity. The software Oufti is required to convert the image data into MATLAB structure array and save in .mat files. Our script serves as an interface containing tools to visualize the characteristics of spatial features such as polarity.
The scripthas been tested on the following systems:
- Windows: Microsoft Windows 10 Home (1903)
- Linux: Centos Linux 7 (core)
For exporting .mat files from microscopy data, Oufti is essential for users to obtain information of single bacterial cells. The software is free to download from https://oufti.org/download/register.php in which minimal MATLAB runtime is provided. To understand more uses based on Oufti, please visit https://oufti.org/tutorial.htm.
The scientific-related packages are imported and applied by our analysis scripts.
numpy
scipy
pandas
matplotlib
seaborn
git clone https://github.com/dwgoblue/SynPolarityAnalysis.git
You need to follow the following workflow to ensure correct analysis.
- Load phase contrast images into Oufti as channel 0.
- Load fluorescence signals with unipolar protein, e.g. PopZ, as channel 1.
- Load fluorescence signals into channel 2 from the proteins with or without the controls of the unipolar protein, such as SpmX and DivIVA in our paper.
Once finished, cells could be isolated manually or automatically which initiates a structure array to record information of each cell. The next step is to reuse meshes which records the positions of cells in order to extract fluorescent signals. We recommend visit the Oufti instructions (https://oufti.org/application.htm) and get more details about the uses of Oufti.
The procedure to analyze time-lapse images is similar to the process for snapshot images. Specifically, the recordings have to be separated by time points. It is noted that the images can only contain one cell and its daughter cell if you plan to analyze single cell division e.g. kymograph. More importantly, one folder must contain one cell data, and the name of the .mat files must stick to the series of time points. For example, time-lapse .mat files of a cell with ID:1 have to be put in the same folder and. The files inside the folder are named 1, 2, 3 if the files are recorded at three sequential time points.
.
+-- [Jul72020]TimeLapseData
+-- Cell1
| +-- 1.mat
| +-- 2.mat
| +-- 3.mat
+-- Cell2
| +-- 1.mat
| +-- 2.mat
| +-- 3.mat
...
There are two ways that you can inform the program where the files are for analysis:
- You are able to add a new function block in
DataInput.pyfor which here provides a template.
def fig1a():
parent_path = r'you_path'
# If every channel is separated, np.array(['TwoInOne']) should be added in the list.
# If there is no PopZ or other controller protein in channel one, np.array(['NP']) should be added in the list.
group1 = [np.array(['TwoInOne']), np.array(['NP'])]
group2 = [np.array(['TwoInOne'])]
# File names are named by their ID.
for mark in ['01', '02', '03']:
img = np.array(['CTRL-{}.mat'.format(mark)])
group1.append(img)
for mark in ['01', '02', '03']:
img = np.array(['EXP-{}.mat'.format(mark)])
group2.append(img)
# Name of each group.
labels = np.array(['CTRL', 'EXP'])
Data = {}
Data['parent_path'] = parent_path
Data['exp'] = exp
Data['labels'] = labels
return Data
- The second way is to create or edit
DataInput.txt. In the file, you are asked to fill out the table like the example below.
=======================================================
parent_path:/home/user/data
input_mode:file
group_number:3
channels:2;2;2
PopZ:Y;Y;Y
labels:Red;Blue;Green
name_1:Red1.mat;Red2.mat
name_2:Blue1.mat;Blue2.mat
name_3:Green1.mat;Green2.mat
=======================================================
parent_path: The script should be able to access the folder of "all" data by the path. Absolute path is recommended.input_mode: The type of inputs inname_Ncan be files or a folder. That is, you can enterfolderin this field instead.group_number: Please write down how many groups are used for analysis which should correspond to the number ofname_N.channels: It depends on how many channels of fluorescence are recorded in your.matfiles. Please typePopZ: To inform the script whether the group has the expression of the unipolar protein PopZ,Y(YES) orN(NO) is entered into this field. It is also applicable if you are using other protein with similar behaviors.labels: Name of each group. You may want to show a concise name for an experimental group on figures.name_N: Exact names of files or folders. You may want to add a newname_Nfield if there are more groups. Noted thatNshould be strictly follow the sequence of numbers.
The next section will introduce how to execute the analysis by these two strategies.
The script offers four pipelines to analyze microscopy data converted from Oufti, including
PopZOnlyPipline
DiffuseSignalPipline
TwoChannelPipline
TimeLapsePipline
If there is only one channel which is PopZ signal saved in the .mat file, we can apply the PopZOnlyPipline in which tot (total intensity) and pks (peak score) analysis are offered. Here is an example to call the function.
from Piplines import*
Starter = PiplinesObj(fig2j,
TimeLapse=False,
columns=[(0,1)])
Starter.TwoChannelPipline(plots=['tot'])
###Signals Recorded from Diffusible Protein
If you are not assumed the protein with spatial patterns, and there is only one channel saved in the .mat file, we can apply the DiffuseSignalPipline in which tot-box (total intensity with boxplot) andtot-box (total intensity with violinplot) analysis are offered. Here is an example to call the function.
from Piplines import*
Starter = PiplinesObj(fig2j,
TimeLapse=False,
columns=[(0,1)])
Starter.DiffuseSignalPipline(plots=['tot-box'])
If you are not assumed the protein with spatial patterns, and there is only one channel saved in the .mat file, we can apply the DiffuseSignalPipline in which tot-box (total intensity with boxplot) andtot-box (total intensity with violinplot) analysis are offered. Here is an example to call the function.
from Piplines import*
Starter = PiplinesObj(fig2j,
TimeLapse=False,
columns=[(0,1)])
Starter.TwoChannelPipline(plots=['tot'])
To analyze the linescans of two profiles, however, it requires an additional input which is the labels of the select groups.
Starter.TwoChannelPipline(plots=['twoprofiles'],
twoselect=['NSSCPG', 'NSSCG'])
Because Oufti only offers two channels for analysis, you may want to combine two files together if there are three types of fluorescence. For this scenario, we need to ask the function to analyze two files shared the same PopZ (or other controller protein) channel. Thus, the additional input contains the indices of the files that should be bound together.
Starter.TwoChannelPipline(plots=['pearson'],
FPextension=[1, 2])
Moreover, the function offers choices tot, pearson, ps, lr, twoprofile, ms, trace and statprofile for analysis.
After rearranging files and folders following to the rules mentioned in the previous section, you are able to execute the analysis for time-lapse data.
from Piplines import*
Starter = PiplinesObj(TL_Exp, TimeLapse=True)
Starter.TimeLapsePipline()
If you choose to introduce the input files via the DataInput.txt, the following code block is recommended. Noted that the path fed into the function indicates the location of the .txt file.
from Piplines import*
Starter = PiplinesObj(TL_Exp, TimeLapse=True)
Starter.TimeLapsePipline()
In this section, we will go through an example analysis process step by step. Hopefully, the information could provide a guidance of how to use the script.
While the location of data could be change, the relative path of each script must be maintained the same as the structure below.
.root
+-- PlotFunc.py
+-- Piplines.py
+-- ReadDataFunc.py
+-- AnalysisFunc.py
+-- DataInput.py
+-- AdditionalFunc.py
+-- DataInput.txt
+-- Demo_notebook.ipynb
+-- Demo_exe.py
+-- Demo
| +-- [Demo]DataInput_folder.txt
| +-- 32NSSC
| +-- 32NSSC.mat
| +-- 32NSSCP
| +-- 32NSSCP.mat
...
To indicate where the data is and other essential information for running analysis, we have to change the content of the .txt file. Here, we provide an example file in the Demo folder, [Demo]DataInput_folder.txt, in which "folder" mode is implemented.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++
parent_path:./Demo
input_mode:folder
group_number:2
channels:2;2
PopZ:N;Y
labels:NSSC-32;NSSCP-32
name_1:32NSSC
name_2:32NSSCP
+++++++++++++++++++++++++++++++++++++++++++++++++++++++
Because red and green fluorescence were recorded in the files, we entered 2 for the field of channels. However, there was no PopZ expressed in the group of NSSC-32 which should be marked by N in the PopZ field.
Before running the script, we have to modify the code block of the main program in Demo_notebook.ipynb or Demo_exe.py.
from Piplines import*
# Apply absolute path if needed.
path = r'./Demo/[Demo]DataInput_folder.txt'
# Change the first input from _ to "" in older version.
Starter = PiplinesObj(_, TimeLapse=False,
columns=[(0,1)],
mode='txt',
txt_path=path)
# Analysis of LR ratio and two-profiles comparison.
Starter.TwoChannelPipline(plots=['lr', 'twoprofiles'],
twoselect=['NSSCP-32', 'NSSC-32'])
If Demo_notebook.ipynb is chosen to run the demo, please initiate Jupyterlab or Jupyter notebook. After that, the code block of analysis is capable of running.
(Terminal)$Jupyter-lab
Alternatively, you could simply execute the analysis with Demo_exe.py. In the terminal, you can run the script by the command below.
(Terminal)$cd [the path to access the Demo folder]
(Terminal)$python3 ./Demo_exe.py
If it works, the output should be
Start to process the group of NSSC-32 .
Processing 30 cells in signal0.
Processing 30 cells in signal1.
Start to process the group of NSSCP-32 .
Processing 29 cells in signal0.
Processing 29 cells in signal1.
The group NSSC-32 has n= 30 after unipolar PopZ selection.
The group NSSC-32 has n= 30
The group NSSCP-32 has n= 15 after unipolar PopZ selection.
The group NSSCP-32 has n= 15
ttest_ind: t = -6.59319 p = 5.01037e-08
The analysis is finishing.
Finally, the figures are saving in the ./Demo folder which should be the same as the results below.
Figure 1. The ratio between two poles.
Figure 2. The comparison between two profiles.
This project is covered under the Apache 2.0 License.