Tuesday, 26 June 2007 ===================== Present: - Eric Deutsch - Jim Schofstahl - Puneet Souda - Luisa Montecchi-Palazzi - Lennart Martens Further CV discussion --------------------- - Pete and Luisa discuss it internally first - In the CV, anything that is a child 'Object Attribute' will need a value. Any value that is also a number (float or int) should also have a unit (from the same CV). This is possible since even (e.g.) intensity has a unit in the PSI-MS CV (in casu: counts, ...) MRM instance document --------------------- - Eric is revising the MRM example in accordance with the conclusions reached yesterday. - During the revision, CV usage of the instance was updated/discussed as well, and where necessary, terms where added. Minor schema quibbles --------------------- - Lennart is collating these for easier discussion. - Changes already done (on Monday) + SpectrumHeader --> SpectrumDescription + SpectrumDescription - InstrumentSetting --> Scan + 'msLevel' attribute added to Scan - Changes done on Tuesday: + Addition of many CV terms + Added 'fileContentTypes' tag under 'admin' in response to the questions by Jimmy Eng. Essentially a placeholder for cvParams. + Allowed two optional attributes under cvParam for values that require a unit. Right now, the units do need to come from the same CV as the term, since the 'cvLabel' is shared. This should not be too much of an issue, since we will put the (rather limited) range of the required units. + Software will contain a special element, called 'softwareParam' with the following, attributes: 'cvLabel', 'accession', 'name', 'version'. + Data processing: a cv param NOT given here is considered 'false', one that is here and has value 'false' is also 'false, one that is here and has NO value, is 'true', and one that is here and is 'true' is 'true'. + Instrument is no longer a component, but rather an extension to ParamGroup. It also now should contain 3 or more components (implied is one of each: source, mass analyzer and detector). + A component can no longer have any child components, to stop recursion on the analyzer or the source for instance. + DataGroupList was deleted. + DataArray and BinaryDataArray were merged in BinaryDataArray + Instrument stuff: - We will have a component list (so it's ordered) and we may want to interleave source/analyzer (say dual source like the LTQ +EDT or secondary source in TOF/TOF) - two options: * go for CV only (so list of extended cvParams (say we call them 'component') that have an 'order' and a 'type'). * use inheritance WITHOUT a top-level 'element' (like before in dataXML); so we need to define three types: 'source', 'analyzer' and 'detector' and have them extend a common supertype, called 'component' (a beefed-up cvParam with an order) and have them be a substitutionGroup. This would have to work with JAXB. - Things still to be discussed: + Mechanism to reference from analysisXML to mzML: can they point uniquely to a single spectrum (URI (file pointer) + spectrum ID)? + The requirement level of attributes and elements has not been specified everywhere + Run List: do we want one run or multiple? Probably we will keep multiple, but it causes some issues with the index, and could lead to very, very, very big files. + AcquisitionList and Acquisition: what is it for, what does it do (primarily for the spec doc) CVS for the files created ------------------------- - Luisa and Lennart will take care of that. (See todo's inn separate document)