Exploring Williams-Beuren syndrome using myGrid.

MOTIVATION: In silico experiments necessitate the virtual organization of people, data, tools and machines. The scientific process also necessitates an awareness of the experience base, both of personal data as well as the wider context of work. The management of all these data and the co-ordination of resources to manage such virtual organizations and the data surrounding them needs significant computational infra-structure support. RESULTS: In this paper, we show that (my)Grid, middleware for the Semantic Grid, enables biologists to perform and manage in silico experiments, then explore and exploit the results of their experiments. We demonstrate (my)Grid in the context of a series of bioinformatics experiments focused on a 1.5 Mb region on chromosome 7 which is deleted in Williams-Beuren syndrome (WBS). Due to the highly repetitive nature of sequence flanking/in the WBS critical region (WBSCR), sequencing of the region is incomplete leaving documented gaps in the released sequence. (my)Grid was used in a series of experiments to find newly sequenced human genomic DNA clones that extended into these 'gap' regions in order to produce a complete and accurate map of the WBSCR. Once placed in this region, these DNA sequences were analysed with a battery of prediction tools in order to locate putative genes and regulatory elements possibly implicated in the disorder. Finally, any genes discovered were submitted to a range of standard bioinformatics tools for their characterization. We report how (my)Grid has been used to create workflows for these in silico experiments, run those workflows regularly and notify the biologist when new DNA and genes are discovered. The (my)Grid services collect and co-ordinate data inputs and outputs for the experiment, as well as much provenance information about the performance of experiments on WBS. AVAILABILITY: The (my)Grid software is available via http://www.mygrid.org.uk

The (my)Grid ontology: bioinformatics service discovery.

(my)Grid supports in silico experiments in the life sciences, enabling the design and enactment of workflows as well as providing components to assist service discovery, data and metadata management. The (my)Grid ontology is one component in a larger semantic discovery framework for the identification of the highly distributed and heterogeneous bioinformatics services in the public domain. From an initial model of formal OWL-DL semantics throughout, we now adopt a spectrum of expressivity and reasoning for different tasks in service annotation and discovery. Here, we discuss the development and use of the (my)Grid ontology and our experiences in semantic service discovery.

Ontology driven dynamic linking of biology resources.

Biologists were early adopters of the Web and continue to use it as the primary means of delivering data, tools and knowledge to their community. The Web is made by the links between pages, yet these links have many limitations: they are static and maintained by hand; they can only link one lexical item to another single resource; ownership is necessary for the placement of link anchors and the link mechanism is essentially inflexible. Dynamic linking services, supported by ontologies, offer a mechanism to overcome such restrictions. The Conceptual Open Hypermedia Service (COHSE) system enhances web resources through the dynamic addition of hypertext links. These links are derived through the use of an ontology and associated lexicon along with a mapping from concepts to possible link targets. We describe an application of COHSE to Bioinformatics, using the Gene Ontology (GO) as an ontology and associated keyword mappings and GO associations as link targets. The resulting demonstrator (referred to here as GOHSE) provides both glossary functionality and the possibility of building knowledge based hypertext structures linking bioinformatics resources.

Variation in airways resistance when defined over different ranges of airflows.

In eight normal subjects airways resistance (Raw) was assessed over six ranges of airflow (about zero flow at both minimum and maximum lung volumes) using an automated whole body plethysmograph. The intervals of flow used were 21s-1 and 11s-1 spanning zero flow, and 11s-1 and 0.51s-1 measured up to and from zero flow. The wider intervals gave less variable results, the coefficients of variation being of the order 11%, 15%, and 22% for the 2, 1, and 0.51s-1 intervals respectively. In all subjects, at minimum volume Raw was some 1.5 times greater when measured over the ranges at end-expiration than at start-expiration; at maximum volume Raw was some 1.3 times greater when measured over the ranges at end-inspiration than at start-expiration. A slight increase in the slopes of the oscilloscope traces used to determine Raw was observed at minimum volume compared with those at maximum volume. In view of the substantial differences reported it is essential that the exact range of flow, and the respiratory phase used, is described when reporting measurements of Raw. Least variability is obtained by estimating slopes over a wide interval of flow, such as 1 or 21s-1.

A comparison of manual and automated methods of measuring airway resistance and thoracic gas volume.

An automated method of measuring airways resistance (Raw) and thoracic gas volume (Vtg) by whole body plethysmography using a digital computer is described. The results obtained by this method are compared with those obtained by an experienced observer and the results of consistency tests on the two methods are given. There was wide variability in the values of Raw obtained using each method, and no significant differences were found between the two methods. The values of Vtg obtained by them were highly correlated, but the results obtained manually were higher than those obtained automatically by an amount proportional to the volume. The consistency of the computer measurements was shown to be greater than that of the manual method. A significant decrease between successive measurements of Raw in some subjects was noted in the manual observations but not in those derived from the automated method. The new method is suitable for the rapid determination of Raw and Vtg.

CINEMA-MX: a modular multiple alignment editor.

UNLABELLED: Analyzing and visualizing multiple sequence alignments is a common task in many areas of molecular biology and bioinformatics. Many tools exist for this purpose, but are not easily customizable for specific in-house uses. Here we report the development of an editor, CINEMA-MX, that addresses these issues. CINEMA-MX is highly modular and configurable, and we present examples to illustrate its extensibility. AVAILABILITY: The program and full source code, which are available from http://www.bioinf.man.ac.uk/dbbrowser/cinema-mx, are being released under a combination of the LGPL and GPL, for Unix or Windows platforms.

Variation between observers in the estimation of airway resistance and thoracic gas volume.

Variations in the estimation of airway resistance (Raw) and thoracic gas volume (Vtg) made by different observers using body plethysmography have been investigated. Five observers determined Raw and Vtg in normal subjects using prerecorded signals so that intrasubject variations were eliminated. The slopes of the oscillography tracings required to make the determinations were assessed using a cursor and a scale fitted to the oscilloscope. Significant and consistent differences in Raw determined by different observers were found,with a mean range of 37%. Great care should therefore be taken when comparing results obtained by different observers. The variation between observers measuring Vtg was significant and consistent but of much smaller magnitude than was the case for Raw, having a mean range of 12%.

A difference in the composition of bronchial mucus between smokers and non-smokers.

The ratio of the amount of sulphated to sialic acid mucin (Su/Si) in the mucous glands of the human tracheobronchial tree has been investigated in seven smokers and seven non-smokers. The two mucins were studied in histological sections stained by the high iron diamine/Alcian blue pH 2-5 sequence and assessed by a point-counting method. Su/Si was greater in the smokers than in the non-smokers, who were almost completely distinguished by this ratio. A decrease in the ratio with each generation of branching from the trachea down the inferior lingular bronchial segmental pathway was seen in both the smoking and non-smoking groups. Analysis of the logarithm of Su/Si showed the smoking group means to be 2-3 times that of the non-smokers at each generation, and over both groups the average decrease down successive generations was given by a factor of 0-9.

Distribution of acid mucus in the bronchial mucous glands.

The percentage area of the bronchial glands taking up the Alcian blue (pH 2-6) stain for acid mucin was obtained from sections from the trachea and from each generation down the inferior lingular axial bronchial pathway in five non-smokers and five cigarette smokers. The mean percentage value for the smokers (59-5%) was significantly less than for the non-smokers (74-4%). The percentage increased between successive generations from the trachea by an average of 6-5. This gradient did not differ significantly between the smoking and non-smoking groups. It was not possible to attribute these differences to any specific mechanism because many interacting variables were involved which could not be measured. Nevertheless the percentage mucin retention or storage provides an introduction to consideration of these variables and their response to cigarette smoke.

A comparative study of the accuracy of area assessment by point counting for bronchial mucous glands.

The accuracy with which areas of bronchial mucous gland on histological sections may be determined has been investigated with respect to inherent statistical variability in the point counting procedure used, and with respect to variations between the histological stains and observers employed. Very good argeement was obtained between areas determined by point counting and by planimetry. It was shown empirically that for sections of well-defined gland-like structures the accuracy (coefficient of area) of area determination was inversely proportional to the 3/4 power of the mean point count. The constant of proportionality depended on the structure's shape and on the point geometry of the grid used. Using the relationship the count needed to achieve a required accuracy of area determination could be established. It was shown that in general a smaller count was required for a given accuracy than would be the case using randomly dispersed tissue. On histological sections the accuracy of area determination is also dependent upon gland boundary definition, and our experiments showed that a recticulin Alcian Blue stain best defined mucous gland acini, and that a PBS stain gave the most accurate results for whole gland. These experiments also showed that there were small but significant differences in mean areas determined by different observers. The observers were shown to be equally consistent in their judgment of what constituted whole mucous gland and acini, but since they differed over mean counts it is recommended that studies should be designed to use either a single observer or to assign observers randomly among groups. With these considerations point counting can be made an accurate method of area determination for these or similar tissues.

Bronchial gland histochemistry in lungs removed for cancer.

The bronchial glands in the main bronchus and succeeding generations of the inferior lingular airway have ben studied in 10 left lungs removed surgically for lung cancer. The ratios of sulphated to sialidated mucin found did not coincide with previous results and suggested that this is not a useful measurement for the assessment of cigarette smoke or other inhaled pollutants. The total acid mucin, measured as a percentage of bronchial gland present, showed the same trends as previously reported--namely, more storage of mucin in non-smokers and in more distal generations. The large dose of cigarettes smoked by the patients in the present series was reflected by a lower level of acid mucin storage, and this measure is potentially useful for assessing exposure to cigarette smoke and other inhaled pollutants. The percentages of mucous and serous cells in successive generations of an airway are recorded for the first time in these smokers and in a "normal" postmortem specimen. The results suggest that similar proportions of mucous and serous cells exist in both large and small airways, but with a tendency in the smokers for more mucous cells in the more distal generations.

Semantic similarity measures as tools for exploring the gene ontology.

Many bioinformatics resources hold data in the form of sequences. Often this sequence data is associated with a large amount of annotation. In many cases this data has been hard to model, and has been represented as scientific natural language, which is not readily computationally amenable. The development of the Gene Ontology provides us with a more accessible representation of some of this data. However it is not clear how this data can best be searched, or queried. Recently we have adapted information content based measures for use with the Gene Ontology (GO). In this paper we present detailed investigation of the properties of these measures, and examine various properties of GO, which may have implications for its future design.

Investigating semantic similarity measures across the Gene Ontology: the relationship between sequence and annotation.

MOTIVATION: Many bioinformatics data resources not only hold data in the form of sequences, but also as annotation. In the majority of cases, annotation is written as scientific natural language: this is suitable for humans, but not particularly useful for machine processing. Ontologies offer a mechanism by which knowledge can be represented in a form capable of such processing. In this paper we investigate the use of ontological annotation to measure the similarities in knowledge content or 'semantic similarity' between entries in a data resource. These allow a bioinformatician to perform a similarity measure over annotation in an analogous manner to those performed over sequences. A measure of semantic similarity for the knowledge component of bioinformatics resources should afford a biologist a new tool in their repertoire of analyses. RESULTS: We present the results from experiments that investigate the validity of using semantic similarity by comparison with sequence similarity. We show a simple extension that enables a semantic search of the knowledge held within sequence databases. AVAILABILITY: Software available from http://www.russet.org.uk.