TDWG Phylogenetics Standards Workshop

The workshop took place on Oct 19, 2008, at the 2008 TDWG Annual Meeting in Fremantle (Perth),

Table of Contents:

Agenda

Workshop objectives, agenda, and initial participant list were published in advance and disseminated to all TDWG conference attendees 1 week prior to the event.

Workshop Notes

Attendees: 18 in total

Introductions

• Stan Blum: TDWG process
  • Foster sharing of standards, data, and biodiversity informatics developments
  • Process is fairly lightweight
  • Put up (social) interfaces for other people to discover you and find out how to contribute
  • Interest group is a group of people that have a shared set of problems that they would like to address.
  • Key piece for a group to "declare itself" is a charter: what is it that the group aims to do.
  • Then create Task Groups, each of which will also have a charter that says what the Task Group will do.
  • Standards are essentially documents, specifications.
  • Infrastructure in the form of mailing lists and wikis are provided.
  • Ratification is initiated by the convener of the Task Group
    • Submits the specification document to the Executive Committee
    • Review manager is assigned and arranges review of the specification
    • Public review follows after review recommendations are taken care of
    • There is no voting process.
  • Q: TDWG is reported to be very slow.
    • Key to keep pushing on the committees, communication is key
    • Process being in place helps in comparison to previously annual pace
  • Q: how is participating in TDWG better than just putting stuff up in SourceForge??
    • There's a lot of experience among TDWG that is related to many of data and types of data and problems we would be dealing with.
    • Can take advantage of the participants/experts in other groups.
    • Interface and communication between biologists and information technologists, which is what TDWG is well set up for
  • Q: what is a good (or a bad) charter?
    • Charters and process have only been instituted since 2 years. The experience and lessons learned from this are limited at this point.
    • Charters ideally are updated at least once annually.

Session I

• David Kidd:
  • proper representation of geographical areas
  • position of observed and inferred nodes, branch paths
  • vicariance-related metadata
  • branches can be segmented, using different methods, representing for example time, or shortest distance (which in projections isn't necessarily a straight line)
  • paleocontinental reconstruction methods and/or parameters and simulation metadata needed
  • data from stratigraphy can have (dating) errors associated with it
  • Pyramid of standards: space (geographic, GML), place (ecological, EML), time (stratigraphic), form (taxonomic): shouldn't phylogenetic standard be in the middle of all of this
• Rick Ree:
  • where does X (extant, or ancestral taxon) live?
    • where has it been found or collected?
    • expert opinion (monographs, floras)
  • describing the geographic range
    • quantitative: lat/long, grid cell values
    • qualitative: geopolitical units
    • predictive: ecological niche, model?
  • inspiration from use cases: historical biogeography, ancestral range estimation
  • take advantage of analogy from standards and ontologies developed for characters?
    • geographic range as an emergent trait of a taxon
    • take standards that have already been developed (OGS geospatial standards) and look at them through the lens of phylogenetics)
    • TDWG work can consist of recommending certain ways to applies an external standard
    • One of the first questions could be to determine whether we can exchange the biogeographical data and reconstructions (e.g. from DIVA and LAGRANGE) that we already have
• Greg Riccardi
  • 230,000 images at present, several TBs of space
  • Capture and track the data that support phylogenetic inference based on characters
  • Morphbank objects (images, collections, etc) have external links: specimen, sequence
    • Chain of evidence
  • Morphbank is being used by external tools, for example MX, as the underlying image store
  • Character state annotation: "sort a bale of plants" metaphor for images
  • Character definitions in trees data files are typically much to short and limited to search databases such as Morphbank
  • Linking images to anatomy ontology terms: ideally have an outline of the part linked, not just the whole image or a pointer within the image
  • Defining characters and states by using ontology terms: capturing, and linking from states
    • would also enable the ability to infer the relatedness of characters
  • Q: Can we have ontologies that are based in phylogeny. It is impossible otherwise to simply combine different morphological ontologies.
  • Q: what is the role of ontologies to informatics standards? -> metadata (property) meaning standardization
  • There is no good way currently to link various annotations in various media about a digital or collection object

Session II

• Rutger Vos: NeXML format
• Chris Zmasek: PhyloXML, phylogenomics
  • using phylogenies for functional inference
  • Q: library support? -> Forester, BioPerl
• Hilmar Lapp: PhyloDB, BioSQL, PhyloWS
• Nico Cellinese: Phylogenetic nomenclature
  • Define names not based on organismal traits but based on phylogenetic relationships

Breakout Groups:

1. Phyloinformatic Web services (Bill P., Rutger, Cindy)
   • data services. which data or metadata are needed from providers
   • data demands that ask for portals (such as GBIF, EOL)
   • crosstalk & provenance between providers
   • scope recommendations, workflows, use cases
2. Metadata standardization & Ontology (David, Chris, Peter, Aaron, Bill, James)
   • metadata uses, properties, semantics
   • reuse possibilities for other TDWG standards
   • expressing the domain model independent of technology
3. Deposition to repositories
   • incentives and standards to increase deposition rates
   • reporting requirements to enable repurposing

Report-out from the groups

Group 1) Bill P.

• divided tasks between "tree decoration" and "tree delivery"
• tree decoration:
  • coevolution (food web, pollinator, host-parasite): for a given set of hosts, give me the tree of parasites
  • computational: calculate divergence times or ages
  • use the tree as input, get it out decorated with certain metadata
• tree delivery
  • types are parameter query and computational query
  • results in ID list
  • given an ID, the desired view (what elements are to be returned) and format, return the object(s)
  • ability to say, give me all trees that contain human, but only those that are about apes
    • ability to give a scope of the desired trees
• ability to dump data, be alerted to updates (e.g., RSS feed)
• different levels of hierarchy of objects: analyses, matrices, trees

Group 2+3) David

• attributes about what makes up a phylogeny, provenance, where data came from and what type, papers possibly associated, parameters used, when and where was it done
• branch length, support values, need multiple of these
• tree to tree relationships
• breaking branches into segments
• sets of nodes, within and between trees, and sets of trees, relationship between nodes (e.g., homologs, host-parasite)
• attributes of nodes (type of node, species or gene, taxon concept, area, more than object per node)
• gene trees are a big application of trees
• talk to geospatial group to learn about their objects and standards
• talk to technical architecture group to learn more about ontologies
• started with an exercise that needs to be followed up with
• there seems to be a taskgroup developing ontologies for this area
• formulating the elements and use cases would create momentum

Interest Group charter

• Keep it simple
• encouragement to list core members
• implementation is the proof, further those efforts
• intersection, don't over-think, what are the core elements

-- HilmarLapp - 31 Oct 2008