Restriction landmark genomic scanning (RLGS) spot identification by second generation virtual RLGS in multiple genomes with multiple enzyme combinations.

BACKGROUND: Restriction landmark genomic scanning (RLGS) is one of the most successfully applied methods for the identification of aberrant CpG island hypermethylation in cancer, as well as the identification of tissue specific methylation of CpG islands. However, a limitation to the utility of this method has been the ability to assign specific genomic sequences to RLGS spots, a process commonly referred to as "RLGS spot cloning." RESULTS: We report the development of a virtual RLGS method (vRLGS) that allows for RLGS spot identification in any sequenced genome and with any enzyme combination. We report significant improvements in predicting DNA fragment migration patterns by incorporating sequence information into the migration models, and demonstrate a median Euclidian distance between actual and predicted spot migration of 0.18 centimeters for the most complex human RLGS pattern. We report the confirmed identification of 795 human and 530 mouse RLGS spots for the most commonly used enzyme combinations. We also developed a method to filter the virtual spots to reduce the number of extra spots seen on a virtual profile for both the mouse and human genomes. We demonstrate use of this filter to simplify spot cloning and to assist in the identification of spots exhibiting tissue-specific methylation. CONCLUSION: The new vRLGS system reported here is highly robust for the identification of novel RLGS spots. The migration models developed are not specific to the genome being studied or the enzyme combination being used, making this tool broadly applicable. The identification of hundreds of mouse and human RLGS spot loci confirms the strong bias of RLGS studies to focus on CpG islands and provides a valuable resource to rapidly study their methylation.

Contour area filtering of two-dimensional electrophoresis images.

We describe an algorithm, Contour Area Filtering, for separating background from foreground in gray scale images. The algorithm is based on the area contained within gray scale contour lines. It can be viewed as a form of local thresholding, or as a seed growing algorithm, or as a type of watershed segmentation. The most important feature of the algorithm is that it uses object area to determine the segmentation. Thus, it is relatively impervious to brightness and contrast variations across an image or between different images. Contour Area Filtering was designed specifically for image analysis of 2D electrophoresis gels, although it can be applied to other gray scale images. A typical gel image is an electrophoretogram or a phosphor image of 1000-2500 spots representing protein or DNA restriction fragments. The images are quantitative with spot intensities reflective of the number of proteins or the DNA fragment copy number. The background intensity can vary widely across the image caused both by variation in spot density and by the physical laboratory process of creating a gel. Analyzing and comparing gel images entails extracting and segmenting spots, registering images and matching spots, and measuring differences between spots. We present experimental results which show that Contour Area Filtering is a quick, efficient method for separating electrophoresis gel background from foreground with extremely high accuracy.