MetaboClust: Using interactive time-series cluster analysis to relate metabolomic data with perturbed pathways.

MOTIVATION: Modern analytical techniques such as LC-MS, GC-MS and NMR are increasingly being used to study the underlying dynamics of biological systems by tracking changes in metabolite levels over time. Such techniques are capable of providing information on large numbers of metabolites simultaneously, a feature that is exploited in non-targeted studies. However, since the dynamics of specific metabolites are unlikely to be known a priori this presents an initial subjective challenge as to where the focus of the investigation should be. Whilst a number of feed-forward software tools are available for manipulation of metabolomic data, no tool centralizes on clustering and focus is typically directed by a workflow that is chosen in advance. RESULTS: We present an interactive approach to time-course analyses and a complementary implementation in a software package, MetaboClust. This is presented through the analysis of two LC-MS time-course case studies on plants (Medicago truncatula and Alopecurus myosuroides). We demonstrate a dynamic, user-centric workflow to clustering with intrinsic visual feedback at all stages of analysis. The software is used to apply data correction, generate the time-profiles, perform exploratory statistical analysis and assign tentative metabolite identifications. Clustering is used to group metabolites in an unbiased manner, allowing pathway analysis to score metabolic pathways, based on their overlap with clusters showing interesting trends.

A batch correction method for liquid chromatography-mass spectrometry data that does not depend on quality control samples.

The need for reproducible and comparable results is of increasing importance in non-targeted metabolomic studies, especially when differences between experimental groups are small. Liquid chromatography-mass spectrometry spectra are often acquired batch-wise so that necessary calibrations and cleaning of the instrument can take place. However this may introduce further sources of variation, such as differences in the conditions under which the acquisition of individual batches is performed. Quality control (QC) samples are frequently employed as a means of both judging and correcting this variation. Here we show that the use of QC samples can lead to problems. The non-linearity of the response can result in substantial differences between the recorded intensities of the QCs and experimental samples, making the required adjustment difficult to predict. Furthermore, changes in the response profile between one QC interspersion and the next cannot be accounted for and QC based correction can actually exacerbate the problems by introducing artificial differences. "Background correction" methods utilise all experimental samples to estimate the variation over time rather than relying on the QC samples alone. We compare non-QC correction methods with standard QC correction and demonstrate their success in reducing differences between replicate samples and their potential to highlight differences between experimental groups previously hidden by instrumental variation.

Hadoop neural network for parallel and distributed feature selection.

In this paper, we introduce a theoretical basis for a Hadoop-based neural network for parallel and distributed feature selection in Big Data sets. It is underpinned by an associative memory (binary) neural network which is highly amenable to parallel and distributed processing and fits with the Hadoop paradigm. There are many feature selectors described in the literature which all have various strengths and weaknesses. We present the implementation details of five feature selection algorithms constructed using our artificial neural network framework embedded in Hadoop YARN. Hadoop allows parallel and distributed processing. Each feature selector can be divided into subtasks and the subtasks can then be processed in parallel. Multiple feature selectors can also be processed simultaneously (in parallel) allowing multiple feature selectors to be compared. We identify commonalities among the five features selectors. All can be processed in the framework using a single representation and the overall processing can also be greatly reduced by only processing the common aspects of the feature selectors once and propagating these aspects across all five feature selectors as necessary. This allows the best feature selector and the actual features to select to be identified for large and high dimensional data sets through exploiting the efficiency and flexibility of embedding the binary associative-memory neural network in Hadoop.

Analysis of crystallization data in the Protein Data Bank.

The Protein Data Bank (PDB) is the largest available repository of solved protein structures and contains a wealth of information on successful crystallization. Many centres have used their own experimental data to draw conclusions about proteins and the conditions in which they crystallize. Here, data from the PDB were used to reanalyse some of these results. The most successful crystallization reagents were identified, the link between solution pH and the isoelectric point of the protein was investigated and the possibility of predicting whether a protein will crystallize was explored.

Using isoelectric point to determine the pH for initial protein crystallization trials.

MOTIVATION: The identification of suitable conditions for crystallization is a rate-limiting step in protein structure determination. The pH of an experiment is an important parameter and has the potential to be used in data-mining studies to help reduce the number of crystallization trials required. However, the pH is usually recorded as that of the buffer solution, which can be highly inaccurate. RESULTS: Here, we show that a better estimate of the true pH can be predicted by considering not only the buffer pH but also any other chemicals in the crystallization solution. We use these more accurate pH values to investigate the disputed relationship between the pI of a protein and the pH at which it crystallizes. AVAILABILITY AND IMPLEMENTATION: Data used to generate models are available as Supplementary Material. CONTACT: julie.wilson@york.ac.uk SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A high-throughput colourimetric method for the determination of pH in crystallization screens.

The crystallization of proteins is dependent on the careful control of numerous parameters, one of these being pH. The pH of crystallization is generally reported as that of the buffer; however, the true pH has been found to be as many as four pH units away. Measurement of pH with a meter is time-consuming and requires the reformatting of the crystallization solution. To overcome this, a high-throughput method for pH determination of buffered solutions has been developed with results comparable to those of a pH meter.