I belong to the non-XML skilled folk, but catching up
We are also planning to use XML for data transfer from the PCR machine to an outstanding statistical program for further analysis (and further: analysis is actually done by different people, by statisticians, who do not know much about the beauties of PCR).
We are astonished by your initiation and wish you success. I guess it would be (in the future) the duty of the instrument providers to make it available direct export of the data in XML format. Hope they could use your standards. (Happy to see that BioRad is already active.)
I wish to return to the question of repeats, I see that you have discussed it earlier.
I understand that RDML aims to contain the minimal information and could be extended any time, thanks to XML. Still we think that (optional) description of different replicates (biological, RT, technical) is important. A series of sample name plus all these options would identify the individual samples, even when the sample name for the samples of the same origin are identical.
Thus using identical names for the samples of the same origin should be allowed. It is not only to simplify the (lazy) user´s life but also to minimize risk of (typing) mistakes (we use 384 platform).
We would also like to include other descriptors (grouping) of the samples (eg. experimental/treatment group of individuals) so that transfer of this data (from experimenter to the analyst) could be done along the measurement data. Would it be by some of the “free description” fields of sample annotation or an individual element you could decide?
Outsiders comment: In the help (for those unfamiliar with XML) some of the header names are not used consistently.
At the end of the “Sample annotation file” description you note that “only the headers of “sample name” (Should it be read: “sample_id”?) … are always mandatory”.
Likewise for the “Target annotation file”.
Best regards,
Eva
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
FYI: Please see my post on the 'ID' thread, which draws this issue in to another (separation of data analysis from data generation); there is a general solution, which is the use of a workflow description format to 'wrap up' all the bits (repeats, treatments, protocols, data, analysis, etc.).
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
We typically refer to technical replicates by using the same sample and target name. All data points with the same combination of sample and target are therefore technical replicates.
I think it is a good idea to organize the description of samples to allow them to be used by software. I see two types of descriptions: the ones that put the sample into a specific group (e.g. biological replicates, males or females, patients or controls, ...) and those that add a free value to a sample (e.g. length, age, ...). Groups could be defined at the same level as samples, targets, users and be referred to by Xpath. These groups can then later be used for statistical analyses. Different labels could be distinguished by adding a label type as attribute (e.g. <label name="age"> 29 <\label>).
Is this something we should add in the first release of RDML?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
I belong to the non-XML skilled folk, but catching up
We are also planning to use XML for data transfer from the PCR machine to an outstanding statistical program for further analysis (and further: analysis is actually done by different people, by statisticians, who do not know much about the beauties of PCR).
We are astonished by your initiation and wish you success. I guess it would be (in the future) the duty of the instrument providers to make it available direct export of the data in XML format. Hope they could use your standards. (Happy to see that BioRad is already active.)
I wish to return to the question of repeats, I see that you have discussed it earlier.
I understand that RDML aims to contain the minimal information and could be extended any time, thanks to XML. Still we think that (optional) description of different replicates (biological, RT, technical) is important. A series of sample name plus all these options would identify the individual samples, even when the sample name for the samples of the same origin are identical.
Thus using identical names for the samples of the same origin should be allowed. It is not only to simplify the (lazy) user´s life but also to minimize risk of (typing) mistakes (we use 384 platform).
We would also like to include other descriptors (grouping) of the samples (eg. experimental/treatment group of individuals) so that transfer of this data (from experimenter to the analyst) could be done along the measurement data. Would it be by some of the “free description” fields of sample annotation or an individual element you could decide?
Outsiders comment: In the help (for those unfamiliar with XML) some of the header names are not used consistently.
At the end of the “Sample annotation file” description you note that “only the headers of “sample name” (Should it be read: “sample_id”?) … are always mandatory”.
Likewise for the “Target annotation file”.
Best regards,
Eva
FYI: Please see my post on the 'ID' thread, which draws this issue in to another (separation of data analysis from data generation); there is a general solution, which is the use of a workflow description format to 'wrap up' all the bits (repeats, treatments, protocols, data, analysis, etc.).
We typically refer to technical replicates by using the same sample and target name. All data points with the same combination of sample and target are therefore technical replicates.
I think it is a good idea to organize the description of samples to allow them to be used by software. I see two types of descriptions: the ones that put the sample into a specific group (e.g. biological replicates, males or females, patients or controls, ...) and those that add a free value to a sample (e.g. length, age, ...). Groups could be defined at the same level as samples, targets, users and be referred to by Xpath. These groups can then later be used for statistical analyses. Different labels could be distinguished by adding a label type as attribute (e.g. <label name="age"> 29 <\label>).
Is this something we should add in the first release of RDML?