Robust Microarray Image Processing
Year: 2007
Type: book-chapter
Abstract: In this work we have presented a complete solution for robust, high-throughput, two-color microarray image processing comprising procedures for automatic spot localization, spot quantification and spot quality control. The spot localization algorithm is fully automatic and robust with respect to deviations from perfect spot alignment and contamination. As an input, it requires only the common array design parameters: number of blocks and number of spots in the x and y directions of the array. Although fully automatic, there is no guarantee that it will perform well for any array. Therefore, we offer some interactive tools to repair grid in case if it is erroneous. Robust ratio estimation comprises two steps. First, linear regression filtering is used to identify and remove aberrant pixels, and then more traditional segmentation approaches are applied for final estimation. Using the two-step quantification algorithm, we ensure a unique ratio estimate, which is as robust as estimates based on medians and as precise as estimates based on means. Linear regression filtering relies on the fact that the two color channels are expected to be highly correlated. Any contamination, which is uncorrelated in the two channels, can be easily recognized by the algorithm and removed. For noisy (weakly correlated) data, the filter is transparent for the data. Moreover, in this case, linear regression estimates can be biased. Therefore we apply a spot segmentation step to establish the final estimate. The spot quality algorithm provides a value of spot quality reflecting the level of confidence in the obtained ratio estimate at each spot. The unique spot quality value is derived from a set of ten marginal quality parameters characterizing certain features of the spot. The contribution of each quality parameter in the overall quality is automatically evaluated based on the visual classification of the spots, or using information available from the replicated spots, located on the same array or over a set of replicated arrays. Therefore the developed procedure allows us not only to quantify spot quality, but also to identify different types of spot deficiency occurring in microarray technology. The quality values can be used either directly to flag out some spots with the quality lower than the user-defined threshold, or in the follow-up analysis as a weight controlling the contribution/influence of the obtained ratio estimates. There are many possibilities to advance the developed algorithms. For example, several spot localization parameters (γ, α and β), that are currently fixed in predefined values, can be iteratively adjusted to achieve the highest regularity of the generated grid. To enhance spot quantification, we can envisage more sensitive (than the single-case diagnostics for the linear regression model) algorithms for aberrant pixel detection. These perspectives are facilitated by further standardizing microarray technology, so that images are becoming more regular, and more specific models for spots and arrays can be developed and justified. As it was shown, different features of the spot (intensity, size, circularity, etc.) can be quantitatively characterized. These characteristics, besides ratios, may contain useful information for the follow-up analysis. One possibility to utilize this information is presented in this paper: we used them to derive spot quality values. However, we believe that more sophisticated analytical tools can be applied to use spot information in other applications. Exploration of these possibilities creates an interesting perspective for future developments.
DOI: 10.5772/4961
Authors: Eugene Novikov, Emmanuel Barillot
Institutions: Universiti Malaysia Pahang Al-Sultan Abdullah, Institut Curie
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: gold
OpenAlex ID: W1593728888