Code and data on the processing of the pulsed-field gel electrophoresis images: A matlab script

Here a matlab script was presented for lane tracking and band detection on the pulsed field gel electrophoresis (PFGE) images. It can also be used as a software tool for automatic analysis of PFGE images. The data consist of several MATLAB codes which collectively have the task of lane tracking, band detecting and pattern recognition on the PFGE images. The lane tracking stage is semi-automatic and the band detection stage is fully automatic. Finally, the pattern of lanes that includes number of, location, width and light intensity level of bands was obtained.


a b s t r a c t
Here a matlab script was presented for lane tracking and band detection on the pulsed field gel electrophoresis (PFGE) images. It can also be used as a software tool for automatic analysis of PFGE images. The data consist of several MATLAB codes which collectively have the task of lane tracking, band detecting and pattern recognition on the PFGE images. The lane tracking stage is semiautomatic and the band detection stage is fully automatic. Finally, the pattern of lanes that includes number of, location, width and light intensity level of bands was obtained.

Data
The data consist of several MATLAB codes, which collectively have the task of lane tracking, band detecting and pattern recognition on the pulsed-field gel electrophoresis images. The PFGE is a laboratory technique used by researchers and scientists to produce a DNA fingerprint for a bacterial isolate as a group of the same type of bacteria [2e5]. The images were provided by two types of bacteria, including Acinetobacter [6], Staphylococcus aureus [7] which were attached with the article. All the codes and data needed for this purpose available in the Mendeley data source [1]. In addition, there is a video showing the performance of the data and Matlab script. The flowchart of implemented algorithm as Matlab script was shown in Fig. 1.

Experimental design, materials, and methods
The material used in this dataset includes images and codes. The images were collected using PFGE BIORAD at the Microbiology Laboratory of Kermanshah University of Medical Sciences in "tiff" format. To analyse the images, MATLAB-R2009a [8] on a system with Intel Core -i5 2430M, quad-core processor overclocked at 3.2 GHz with 8GB of RAM clocked at 1600 MHz was used. To evaluate and optimize the codes, a trial version of GelCompar II software was also used.

Algorithm for the lane tracking
A program, "lane_tracking.m", is responsible for lane tracking. First, image was converted to grayscale from RGB format. The desired area of the image including lanes was cropped then was resized to 500 Â 500 pixels. Next, image segmented to sub-images to calculate vertical projection profile (VPP).
Specifications Table   Subject Biomedical Engineering Specific subject area image processing in microbiology and biotechnology Type of data MATLAB code, image, video How data were acquired All source codes written in Matlab software. Data format MATLAB code, JPEG, Mp4 Parameters for data collection All the codes were implemented in MATLAB-R2009a on a system with Intel Core -i5 2430M, quad-core processor overclocked at 3. Value of the Data The provided codes can be used to pulsed-field gel electrophoresis image analysis. The Matlab script will allow microbiologist to molecular subtyping. This approach can be used to automatic lane tracking, band detection and pattern recognization on PFGE images.
To detect local of each lane in the sub-images, local maxima from signal of VPP was detected. Using the matlab codes, "widthfind.m" and "Remov_locmax.m", false-positive local maxima in the background was removed. After detecting the center of lanes in each of sub-images, the centers associated with each lane are given to the function "createFit.m" to fit the patch of lane. In detail, the proposed algorithm is described in Ref. [9]. Fig. 2 depicts the lane tracking.

Algorithm for the band detection
The band detection as one of the stages consists of two phases. First, a matlab code, "bandremovebackgroundm.m", is responsible for removing background noise from lanes image. To this end, the lane image was smoothed by an adaptive median filter on lane images. Then, as a completion step, lower envelope of VPP of lane was subtracted. The source code for this function is given bellow.  In addition, the program "band.m" is responsible for band detection. To this end, after resizing it to 100 Â 500 pixels and calculating the VPP of lane, the location of the bands was detected using the function "peakfind.m". The source code for this step is as follows. Finally, the pattern of lanes was determined by calculating the four parameters, including number of location, width and light intensity level of bands.