2st ISA-TAB workshop Outcome/Summary (to date)

1
2st ISA-TAB workshopOutcome/Summary (to date)
This workshop is funded by the BBSRC Tools and
Resources (WODS, BB/E025080/1), with
contributions from EBI and NERC Bioinformatics
Center
Workshops on Data Standards (WODS) EBI,
Cambridge, UK 16th, 17th and 18th June 2008
2
Monday, 16thReviewing XSLT
3
XSLT issues discussed and solutions

• FuGE extensions status
• ACTION (Andy) to create a portal on FuGE page
  to list all the extensions, their status,
  contacts, links and examples etc
• This can help maximize interactions and advertise
  the status of each extensions (like MIBBI does
  for the checklists)
• E.g. the RNAi group (that is building a FuGE
  extension in this domain) needs to develop
  something to describe a microtiter plate and the
  work on the Array Design can be reused
• If there will be a FuGE-derived MAGE-ML then
  this group could reuse the ADF part or other
  parts
• ACTION (Javier,Helen) explore encoding of
  microplate representation in the datafiles and
  referencing from Assay
• Problem is that we do not have a final decision
  on where there will be a MGEDs extension of FuGE
• (ACTION, Helen) to check with the MAGE list
  MGED board

4
XSLT issues discussed and solutions

• Namespace inconsistency
• Should we have FuGE controlled namespaces?
• The OBO Foundry is considering doing it for
  ontologies
• ACTION (Andy) to ask the list if desiderable
• namespace is a critical issue for xsl processing,
  not so much for other parsing methods
• this is probably a recommendation for those that
  wish to use XSLT for presentation purpose a
  dedicated page will be set up on the ISA-TAB
  website to list such XSLT recommendations ACTION
  (Philippe)
• Annotation overloading
• Descriptions are used as term gathering fields
• We could recommend on the fly term creation
  (collection of term as supplied by users)?
• ACTION (Andy, Ally) add recommendations on FuGE
  wiki (explain the use of FuGE)
• Paper soon out on recommendations to extend FuGE

5
XSLT issues discussed and solutions

• Name attribute optionality
• When this is missing XSLT uses the ID, giving a
  less human readable transformations
• We could recommend that name is used when
  readability is required/preferred ACTION
  (Philippe) to modify XSL templates
• Way to categorize assays is not in FuGE
• How to code technology and endpoint to
  categorize the assay (InvestigationComponent)?
• It can be done implicitly, but would useful to
  have these as explicit objects
• However, as there will not be a FuGE v1.1, work
  around to any issues or needs will be done via
  recommendations ACTION (Andy) to add in FuGE wiki
  
• Reagents info are on ISA-TAB
• Flow cytometry examples have more
  depth/granularity, e.g. all the reagents are
  listed they have coded it via Material
• - Even if FuGE recommends to do it via structured
  Protocol (see Gel-ML)
• ACTION (Andy, Ally) point them to design patters
  on FuGE wiki
• ISA-TAB can (somehow) deal with it via Protocol
  Component field, just added

6
XSLT scripts library next steps

• More FuGE ML files are needed to test the current
  scripts
• ACTION (Ally) to give more example to Philippe
  from Symba
• ACTION (Philippe) to send script to Ally
• ACTION (Philippe) to set up a XSLT page on
  ISA-TAB to post all the scripts
• Then the scripts will be tested with FuGE
  extensions e.g. GelML
• More example to test to evaluate and finalize the
  scripts
• ACTION (Frank, Philippe) collaboratively finalize
  the scripts for GelML
• Final comment on scope of ISA-TAB in relation to
  FuGE
• FuGE or other XMLx are more granular/expressive
• We got to accept that fact that when transform in
  ISA-TAB we will/may loose/compress some of the
  info

7
MAGE-TAB to ISA-TAB converter

• To be done by AE team

8
Tuesday, 17thReviewing the ISA-TAB
9
Investigation File Changes and decisions (1)

• Add Investigation PubMed ID and Investigation
  Publication DOI to Investigation section
• - Only for paper describing across Studies
• Studies section need to be singular (Study)
• Comment point Study header is sufficient to
  separate the sections, no need for have
  start/end repeatable block
• If developers want to add a comment then this
  would be this is a comment thingy.
• - Comment must have as the first char
• - But in Study/Assay by adding a column comment
  (see Table 5 in spec v0.3) for the users
• Create a new section Study Publications where
  we group publications attributes, moving the id,
  description and date under Study Section
• Create a new section Study Design Descriptors

10
Investigation File Changes and decisions (2)

• All fields name are case sensitive
• Edit every field must have first letter upper
  case
• Section headers go all upper case
• To allow easy visualization when imported in
  spreadsheet
• File will be interpreted in a Unicode
• Any subsections within a repeatable block (Study)
  must remain within the block
• - But the order of the subsection within the
  block can vary
• Use the triplet (type, accession, source ref)
  consistently, if ontology/CV is used, if not,
  Name and Type are entered as free text (add
  example in the spec)
• Add Protocol Parameter Name, followed by
  Protocol Parameter Type and Protocol Parameter
  Type Term Accession, Protocol Parameter Type
  Term Source REF

11
Investigation File Changes and decisions (3)

• Correct to allow for multiple values, for
  Protocol Parameter Type/Term/Source and Study
  Design Type/Term/Source triplets
• Add Protocol URI and Protocol Version fields
  in the Study Protocol subsection
• The pointer to external file(s) allows to users
  to provide these in the format they wish
• URI should be resolvable
• Ultimately these requirement are up to the
  implementers similarly to make e.g. mandatory
  other Protocol fields
• Remove Protocol Component Parameter, Instrument
  Component, Software triplets and Processing
  fields
• Add Protocol Component Name, Protocol
  Component Types, Protocol Component Types Term
  Accession, Protocol Component Types Term Source
  REF
• Used for listing, e.g. instruments, software,
  reagents, operator
• Semicolon separator ACTION (Marco) to provide
  examples of options
• Clean up all the field names and make them
  unique by prefix them with the name of the
  section, e.g. Investigation PubMed ID vs Study
  PubMed ID

12
Study/Assay File Recommendations

• The table represents a graph and each edge needs
  to appear at least one, nodes do not need to be
  repeated, e.g.
• Microarray (technology) / Gene expression
  (endpoint) tab
• Document how to represent the case when when 2
  different analysis protocols are applied to the
  same set of data file
• In this case we follow MAGE-TAB by repeating
  vertically the data file names (only, not need to
  repeat the previous columns) followed by a new
  analysis protocol and output data file names
• Factors Value can be referenced in both Study and
  Assay tabs
• But the same value cannot be in both tabs,
  examples to be added

13
Tuesday, 17thTools and implementations
14
Scripts and tools plans

• From ISA-TAB to FuGE ML
• To be done (Phil, Ally wants this -)
• Map the ISAcreator java model to FuGE general
  elements
• Ally to help checking/validating mapping
• From FuGE ML to ISA-TAB
• Work in progrosses XSLT under development
• Philippe, Frank, Ally, Nigel and Andy
• ISA-TAB creation
• ISAcreator (will be open source)
• Other tools from participating systems.
• ISA-TAB validation
• - Common, minimum validation rules/scripts to be
  defined/developed (e.g. structure, case
  sensitive)
• Use part of the ISAcreator configuration as
  library
• -gt Google doc with list of basic rules (to be
  identified when creating the v1 spec)
• -gt The ISAcreator config code will be stripped
  down to the basic rules and posted on the ISA-TAV
  sf site (SVN)
• ISA-TAB and MAGE-TAB
• - Helen and Susanna to talk to MGED, ref to NIH
  grant

15
Wednesday, 18th Next steps and publication plans
16
Release plans

• Release candidate 1, ISA-TAB v1
• Philippe, Marco to edit/add all the agree changes
  in the spec
• -gt done by June 27th
• Dave, Ally, Kieran check and review
• -gt done by July 18th
• All to read and comment/suggest
• -gt wiki pages will be set up on ISA-TAB site to
  facilitate discussion
• -gt all comments received by end of August
• Philippe fix the current ISA-TAB examples to
  reflect new spec
• -gt Release candidate 1, ISA-TAB v1 out by mid
  Sept
• This version will include details on fields in
  Investigation file and the list of fields allowed
  in the Study and Assay files
• - The specific Assay files defined by the
  participating communities, will be listed and new
  can be added, without having to release new
  versions

17
Pending issue

• Reference system for SEND and CDISC (STDM)
• Take this discussion on the ISA-TAB list with
  interested parties, Michael and Steve in
  particular
• Subject ID in STDM should be the same of Source
  Name ISA-TAB (as add another subject ID column?)
  then add the file as external
• Each STDM file has a Study ID, Domain ID, Subject
  ID (2 types of these, probabaly we can use the
  UsubjID) and Idvar (column) and Idvarval (column
  value)

18
Publication and next workshop

• Publication suggested content
• Rationale and use case for ISA-TAB
• History from MAGE-TAB to ISA-TAB
• Present it as format (and interface to a format)
  not a standard
• -gt Describe scripts making it interoperable
  with other formats
• Example of implementations to date
• -gt Tools/systems that have output/input in this
  format
• -gt Also (simply) more real examples from
  communities posted on this site
• Start writing end of this year, to submit early
  next year journal to be decided later
• Next workshop would be a users meeting (in 2009)
• To fix minor issues, recommendations,
  ambiguities, sharing development approaches,
  components etc.