MOTIVATION: Multidimensional scaling (MDS) is a well-known multivariate statistical analysis method used for dimensionality reduction and visualization of similarities and dissimilarities in multidimensional data. The advantage of MDS with respect to singular value decomposition (SVD) based methods such as principal component analysis is its superior fidelity in representing the distance between different instances specially for high-dimensional geometric objects. Here, we investigate the importance of the choice of initial conditions for MDS, and show that SVD is the best choice to initiate MDS. Furthermore, we demonstrate that the use of the first principal components of SVD to initiate the MDS algorithm is more efficient than an iteration through all the principal components. Adding stochasticity to the molecular dynamics simulations typically used for MDS of large datasets, contrary to previous suggestions, likewise does not increase accuracy. Finally, we introduce a k nearest neighbor method to analyze the local structure of the geometric objects and use it to control the quality of the dimensionality reduction. RESULTS: We demonstrate here the, to our knowledge, most efficient and accurate initialization strategy for MDS algorithms, reducing considerably computational load. SVD-based initialization renders MDS methodology much more useful in the analysis of high-dimensional data such as functional genomics datasets.

BACKGROUND: Significant inconsistencies between probe-to-gene annotations between different releases of probe set identifiers by commercial microarray platform solutions have been reported. Such inconsistencies lead to misleading or ambiguous interpretation of published gene expression results. RESULTS: We report here similar inconsistencies in the probe-to-gene annotation of Applied Biosystems AB1700 data, demonstrating that this is not an isolated concern. Moreover, the online information source PANTHER does not provide information required to track such inconsistencies, hence, even correctly annotated datasets, when resubmitted after PANTHER was updated to a new probe-to-gene annotation release, will generate differing results without any feedback on the origin of the change. CONCLUSION: The importance of unequivocal annotation of microarray experiments can not be underestimated. Inconsistencies greatly diminish the usefulness of the technology. Novel methods in the analysis of transcriptome profiles often rely on large disparate datasets stemming from multiple sources. The predictive and analytic power of such approaches rapidly diminishes if only least-common subsets can be used for analysis. We present here the information that needs to be provided together with the raw AB1700 data, and the information required together with the biologic interpretation of such data to avoid inconsistencies and tracking difficulties.