Menu
▾
▴
[Animl-develop] [Fwd: Re: AnIML sequence]
|
From: Prof. R. J. L. <rob...@uw...> - 2006-05-29 18:23:39
|
Maren, I was checking on a sample file generated by Anh Dao called something lik= e ADNIR to see why it was crashing my read routine and realised that it did not have an attribute labelled IDENTIFIER under MeasurementDa= ta. StAX routines fail if you try to read an attribute that is absent! I was using this attribute as the equivalent of the JCAMP-DX ##TITLE=3D l= abel since based on the discussion about SampleSet where the sample could be reference or sample simultaneously in 1 aml document then it would seem that the SampleName label could NOT be used as the title equivalent. Does this make sense and thus should not IDENTIFIER be a required label? Thanks Robert -------- Original Message -------- Subject: Re: AnIML sequence Date: Wed, 24 May 2006 10:36:50 +0200 From: Mar...@wa... To: Prof. Robert John Lancashire <rob...@uw...> CC: Prof. Tony Davies <Ant...@gm...>, Mar...@wa... Hi Robert, > Can you give me an example of what you are thinking of here? > > I think at the time you are creating the document it is known > what it is you have taken the measurement on! Hence you know > what is the sample and what may be blanks or references! The idea is that there could be a case where you store multiple measurements in an AnIML file, and the same physical sample is a referenc= e in one, but a run sample in another. In order to not have to waste a lot of space with redundant sample information, each one is only stored once and can assume different roles. Mit freundlichen Gr=FC=DFen / Best regards Maren. Dr. Maren Fiege Senior Product Manager Waters GmbH Europaallee 27, D-50226 Frechen, Germany Tel.+49 2234 9207 - 0 Fax +49 2234 9207-99 http://www.waters.com/informatics "Prof. Robert John Lancashire" <rob...@uw...> 09.05.2006 13:01 To Mar...@wa... cc "Prof. Tony Davies" <Ant...@gm...> Subject Re: AnIML sequence Maren, I think I get the concept that a sample can be a sample in one experiment and a reference in a different experiment but are you suggesting that within the same AnIML document it could change role. Can you give me an example of what you are thinking of here? I think at the time you are creating the document it is known what it is you have taken the measurement on! Hence you know what is the sample and what may be blanks or references! I continually run across this problem of trying to envisage when the document is being generated but figure that the instrument can generate the spectrum section and the user generate the sample section at a later stage. I have tried 2 different approaches to reading the document, 1 using DOM and the other using StAX. StAX fits exactly with the lazy load concept. I think the lazy load idea is OK and in fact for a spectrum display I could totally IGNORE the sample section altogether and never read anything in there, if all I want is an XY plot. It is only when I want to associate sample info with spectrum info that I run into difficulty unless I store all sample info and then read spectrum info and try to work backwards to do the association having eventually found out which is the sample. bear with me while I get this sorted out... Thanks Robert on 09/05/2006 04:04 AM Mar...@wa... said the following: > Hi Robert, >=20 > catching up on my email... >=20 >> I think this is inefficient and whilst there may be benefits to having >> sample separate to measurement I believe it would be easy enough to >> add an attribute at the sample stage to declare which of the >> samplesets was the actual sample and which was the blank etc. >> The sampleID seems at present to be anything you want to call it >> so if that is to stay the same then a new attribute is needed. >> It could be a simple Boolean (isSample or isReference) and would >> only be required if the sample was part of the measurement. > The samples defined in AnIML can be re-used and may change roles, so yo= u=20 > cannot flag them in the sample definition as you suggest. > You are trying to stream in the AnIML file and keep it in memory=20 > altogether. The idea when designing the format was that you should use=20 > something called "lazy load", meaning that you would only read what you= =20 > need at a given time. This is an entirely different paradigm. >=20 > Mit freundlichen Gr=FC=DFen / Best regards >=20 > Maren.=20 >=20 >=20 > Dr. Maren Fiege > Senior Product Manager >=20 > Waters GmbH > Europaallee 27, D-50226 Frechen, Germany > Tel.+49 2234 9207 - 0 Fax +49 2234 9207-99 >=20 > http://www.waters.com/informatics=20 >=20 >=20 >=20 --=20 Prof. Robert J. Lancashire mailto:rob...@uw... Department of Chemistry Tel (876) 512 3026 University of the West Indies, Kingston 7 Fax (876) 9771835 Mona Campus, JAMAICA. http://wwwchem.uwimona.edu.jm/chrl.html --=20 Prof. Robert J. Lancashire mailto:rob...@uw... Department of Chemistry Tel (876) 512 3026 University of the West Indies, Kingston 7 Fax (876) 9771835 Mona Campus, JAMAICA. http://wwwchem.uwimona.edu.jm/chrl.html |
