Download Latest Version BMC Bioinformatics PDEMIP.zip (158.5 MB)
Email in envelope

Get an email when there's a new version of Automatic Microarray Image Segmentation

Home
Name Modified Size InfoDownloads / Week
BMC Bioinformatics PDEMIP.zip 2016-01-04 158.5 MB
ReadMe.txt 2016-01-04 6.5 kB
Totals: 2 Items   158.5 MB 0
				README 
=============================================================================================

			   AMIP 1.0 release
---------------------------------------------------------------------------------------------


This "Amip" GUI is intended to provide code examples of the methods proposed in: 
"Unsupervised image segmentation for microarray spots with irregular contours and inner holes"
authors: Bogdan Belean, Monica Borda, Jörg Ackermann, Ina Koch and Ovidiu Balacescu
Journal: BMC Bioinformatics, 2015, 16:412
 

PDE were used for grid alignment, whereas autocorrelation driven PDE and k-means clustering 
were used for spot segmentation. The proposed segmentation procedure yields the spots median 
intensity values for the microarray images recorded from Cy3 and Cy5 fluorescent dyes. Based on 
the fold change factor, the differentially expressed genes were determined.


Both images with single spot groups and multiple spot groups (372 and 324 spots/group) can be 
processed. Example of such images were taken from the GEO and SMD databases. e.g., GSM102718 - 
GEO database experiment for single spot group, AT20385 SMD microarray experiment with 324 spots/
/group, AT26409 - SMD microaray experiment with 372 spots/group. 


Contents
--------
The "Amip" graphic unit interface integrates methods for (A) processing single spot group 
microarray images, (B) microarray images with multiple spot groups - 372 spots/ group and (C)
microarray images with multiple spot groups - 324 spots/group. 

Amip.m and amip.fig - build the GUI interface which includes control panel for each type of images
(A), (B) and (C).


------------------------------------------------------------------------------------------------
A) single spot group	
------------------------------------------------------------------------------------------------
preprocess342.m    - performance image enhancement to underline microarray spots.
image_profiles.m   - computes image profiles.
shock_filter.m     - applies shock filter on profiles.
create_grid.m      - performs grid alignment based on the shock filter profiles
shock_filter1D.m   - applies shock filter on rows and columns of spots.
Horizontal_separators.m   - dealineates each pot using a rectangle; the ellipse inscribed
			  within rectangle represents the microarray spot.
find_centre.m	   - computes the mass centre relative to pixel intensity values within a spot 
		  in order to be able to perfrorm computation with results within the GEO 
		  data repository.
Note	Apply preprocess by pressing the process button. Chose the threshold based on the shock
----	filter profiles (e.g. 5000 and 5000 for the image GSM102718). Press spot location 
	detection to compute  the mass centre relative to pixel intensity values within each spot 
        in order to be able to perform a comparison with results within the GEO data repository.
	(e.g. Agilent computes mass centre for spot detection)

!Important	The proposed methods were applied on the microarray images recorded from channel 1
----------	e.g. image_ch1.tif

------------------------------------------------------------------------------------------------
B) Multiple spot group - 372 spots / group	
------------------------------------------------------------------------------------------------
preprocess.m		- performance image enhancement to underline microarray spots.	
find_spotgroups.m	- morphological based spot groups detection
image profile.m		- computes profiles within sub-group
shock_filter.m		- applies shock fiters on profiles
create_grid2.m		- performs grid alignment within sub-group based on the shock filter
			profiles
shock_filter1D.m	- applied shock filter on rows and columns of spots.
Horizontal_separators2.m  - dealineates each spot using a rectangle; the ellipse inscribed
			  within rectangle represents the microarray spot.
mairplot.m		- Matlab function to compute differentially expressed genes.

The following table lists the rotation correction factors within images of BMC article:

SMD Image ID	rotation correction (degrees)
---------------------------------------------
	26409		0
	26415		0.2		
	26425		-0.5				
	26426		0.2
---------------------------------------------



------------------------------------------------------------------------------------------------
B) Multiple spot group - 324 spots / group	
------------------------------------------------------------------------------------------------
preprocess342.m		- performance image enhancement to underline microarray spots.	
find_spotgroups324.m	- morphological based spot groups detection
image profile.m		- computes profiles within sub-group
create_grid324.m	- performs grid alignment within sub-group based on the shock filter
			profiles
shock_filter.m		- applies shock filters on profiles
shock_filter1D.m	- applied shock filter on rows and columns of spots.
Horizontal_separators324.m 	- dealineates each spot using a rectangle; the elipse inscribed
			  within rectangle represents the microarray spot.
mairplot.m		- Matlab function to compute differentially expressed genes.

The following table lists the rotation correction factors within images of BMC article:

SMD Image ID	rotation correction (degrees)
---------------------------------------------
	AT20385		0.7
	AT20391		0.3
	AT20392		0
	AT20392		0.3
---------------------------------------------


Requirements
------------
Note that the code examples require Matlab R2011b or later.
Microsoft Office; The results are written in .XLX files: 
(a) C:\Results\MultSpotGr324.xlsx - line of code amip.m - for the AT20385 images 324 spots/group
(b) C:\Results\MultSpotGr372.xlsx - line of code amip.m  - for the AT20409 images 372 spots/group

----------------------------------------------------------------------------------------------------
! Important:  Change  .xlsx file names from one experiment to another. Xls file write error may occur if otherwise..
! Important:  Change  image names within horizontal_separators324 to point to the dataset:

e.g. Reference = imread('C:\Users\Bogdan\Desktop\ULMIS\DataSet\AT20395\20395_ch2.tif',1); 
     Reference2 = imread('C:\Users\Bogdan\Desktop\ULMIS\DataSet\AT20395\20395_ch1.tif',1);
! Important:  results with uu-regulated spot in case of the dataset 1 and 2 are in folder Results
------------------------------------------------------------------------------------------------------
Source: ReadMe.txt, updated 2016-01-04