Related Experiment Videos
Unsupervised technique for robust target separation and analysis of DNA microarray spots through adaptive pixel
Daniel Bozinov1, Jörg Rahnenführer
1Center for Human Molecular Genetics, University of Nebraska Medical Center, Omaha, NE 68198, USA. dbozinov@unmc.edu
Bioinformatics (Oxford, England)
|June 7, 2002
Summary
A new method using pixel clustering improves gene spot intensity assessment in microarrays, outperforming existing techniques for problematic spots with imperfections or low expression.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Microarray image analysis faces challenges due to gene spot imperfections like irregular contours and artifacts.
- These imperfections lead to inaccurate intensity values and R/G ratios, necessitating improved analytical methods.
Purpose of the Study:
- To introduce a novel method for accurate intensity assessment of gene spots in microarray images.
- To address the limitations of current analytical approaches for imperfect or low-expression gene spots.
Main Methods:
- Developed a new technique based on clustering pixels into foreground and background.
- Implemented two clustering algorithms: k-means and Partitioning Around Medoids (PAM).
- Combined Extractiff (Java) and Pixclust (R) software tools for the implementation.
Main Results:
- The novel method, PX(PAM) and PX(KMEANS), demonstrated superior performance compared to existing methods.
- The approach is highly robust against various artifacts through adaptive partitioning.
- Accurate expression intensity values were achieved, especially for problematic gene spots.
Conclusions:
- The proposed pixel clustering method offers a robust and accurate solution for gene spot intensity assessment in microarrays.
- This technique effectively handles imperfections and low expression, improving data reliability in genomic studies.