Separating Putative Pathogens from Background Contamination with Principal Orthogonal Decomposition: Evidence for Leptospira in the Ugandan Neonatal Septisome

High-throughput sequencing technologies have strongly impacted microbiology, providing a rapid and cost-effective way of generating draft genomes and exploring microbial diversity. However, sequences obtained from impure nucleic acid preparations may contain DNA from sources other than the sample. Those sequence contaminations are a serious concern to the quality of the data used for downstream analysis, causing misassembly of sequence contigs and erroneous conclusions. Therefore, the removal of sequence contaminants is a necessary and required step for all sequencing projects. We developed DeconSeq, a robust framework for the rapid, automated identification and removal of sequence contamination in longer-read datasets ( 150 bp mean read length). DeconSeq is publicly available as standalone and web-based versions. The results can be exported for subsequent analysis, and the databases used for the web-based version are automatically updated on a regular basis. DeconSeq categorizes possible contamination sequences, eliminates redundant hits with higher similarity to non-contaminant genomes, and provides graphical visualizations of the alignment results and classifications. Using DeconSeq, we conducted an analysis of possible human DNA contamination in 202 previously published microbial and viral metagenomes and found possible contamination in 145 (72%) metagenomes with as high as 64% contaminating sequences. This new framework allows scientists to automatically detect and efficiently remove unwanted sequence contamination from their datasets while eliminating critical limitations of current methods. DeconSeq's web interface is simple and user-friendly. The standalone version allows offline analysis and integration into existing data processing pipelines. DeconSeq's results reveal whether the sequencing experiment has succeeded, whether the correct sample was sequenced, and whether the sample contains any sequence contamination from DNA preparation or host. In addition, the analysis of 202 metagenomes demonstrated significant contamination of the non-human associated metagenomes, suggesting that this method is appropriate for screening all metagenomes. DeconSeq is available at http://deconseq.sourceforge.net/.

Modern imaging techniques for probing brain function, including functional Magnetic Resonance Imaging, intrinsic and extrinsic contrast optical imaging, and magnetoencephalography, generate large data sets with complex content. In this paper we develop appropriate techniques of analysis and visualization of such imaging data, in order to separate the signal from the noise, as well as to characterize the signal. The techniques developed fall into the general category of multivariate time series analysis, and in particular we extensively use the multitaper framework of spectral analysis. We develop specific protocols for the analysis of fMRI, optical imaging and MEG data, and illustrate the techniques by applications to real data sets generated by these imaging modalities. In general, the analysis protocols involve two distinct stages: `noise' characterization and suppression, and `signal' characterization and visualization. An important general conclusion of our study is the utility of a frequency-based representation, with short, moving analysis windows to account for non-stationarity in the data. Of particular note are (a) the development of a decomposition technique (`space-frequency singular value decomposition') that is shown to be a useful means of characterizing the image data, and (b) the development of an algorithm, based on multitaper methods, for the removal of approximately periodic physiological artifacts arising from cardiac and respiratory sources.