We have set up a data repository on SourceForge to house the growing number of data files that are worked on daily by new curators and students. The repository uses Subversion (SVN) software to maintain current and all previous versions of each Phenex data file.  Curators “update” the local copies of data files on their computers with the current versions from the repository, add their modifications, and “commit” their modified local copy to the repository at the end of their work session.  We have been using ZigVersion as a friendly graphical client interface to SVN.

This is a great improvement to our system of file management. We previously used a server for file sharing, which limited us to saving only the current version of each file.  Aside from version control, SVN has other nice features such as automatic file merging if a curator checks in a different version of a file (which might happen if the file on the repository has been updated by another user since the curator first updated their local copy).  Curators also utilize commit comments to document the edits made on a data file.

After a summer of intensive development, at our recent Data Roundup we introduced curators to our new phenotype annotation software, Phenex. Phenex is an application for annotating character matrix files with ontology terms. Character states can be annotated using the Entity-Quality syntax for ontologically describing phenotypes. In addition, taxon entries can be annotated with identifiers from a taxonomy ontology. Phenex saves ontology annotations alongside traditional character matrix data using the new NeXML format standard for evolutionary data.

Phenex replaces Phenote as our data curation tool.  Phenote was developed by the Berkeley Bioinformatics and Ontologies Project to facilitate the annotation of biological phenotypes using ontologies.  By creating a custom configuration of Phenote to support our data requirements, we were able to immediately begin annotating phenotypes.  Because Phenote provides a flat table of annotations, it works well for curation of the prominent phenotypes of a few genotypes at a time, such as in model organism mutants.  However, evolutionary data is usually collected in the form of a character by taxon matrix, recording a value for each character for every taxon.  Over time we added several custom interface features to Phenote to better support annotation of evolutionary data.  However curation of data from real publications at our first Data Jamboree, in April 2008, revealed that our initial workflow was cumbersome and error prone.

Like Phenote, Phenex facilitates the construction of annotations using ontology terms.  In Phenex, however, the data model and interface are organized in a format that will be familiar to any evolutionary biologist who has used a character matrix.  The new design addresses several requirements highlighted by user testing at the April Data Jamboree:

• Manage independent lists of Taxa, Characters, and States, tied together via a data matrix.
• Facilitate entry of basic free text matrix data by undergraduate assistants.
• Allow morphology experts to focus on translating character state descriptions into ontology annotations without simultaneously managing unwieldy lists of taxa.
• Easily import data from published NEXUS matrix files.
• Include ontology annotations within a standard evolutionary data matrix file format.

We addressed this last requirement by adopting the new NeXML file format.  Being an XML schema, NeXML allows us to embed ontological annotations directly within standard evolutionary data.  Thus the standard data in Phenex documents should be usable in any evolutionary analysis software which supports NeXML.

Although Phenex began as an offshoot of Phenote, it is now a completely separate application hosted at the OBO project site.  Like Phenote, Phenex builds upon OBO-Edit as an application framework for working with ontologies.  While Phenex has been developed with the needs of Phenoscape in mind, it should be useful to anyone working with ontologies and evolutionary data.  We hope it can become a general tool for evolutionary biologists incorporating ontologies into their work.  More information about Phenex can be found on its homepage, including instructions for downloading the application and obtaining its source code.

“…where the buffalo roam and the data are rounded up-up all day….”

A few weeks ago, from Sep 27 to Oct 1, we met in the Black Hills of South Dakota with a group of guest data curators and outside advisors to curate high priority papers, refine the curation workflow and Phenex interface, and evaluate the first prototypes for the web-based user interface to the database. Not only did the workshop end up highly productive (see below), we also had a chance to observe the annual roundup of the largest herd of buffalo in North America, swim in cold Sylvan lake, and see Mount Rushmore one evening.

After participant introductions, the day began with several presentations focused on the utility of synthetic research databases (Monte Westerfield) and the way that phenotype-genotype relationships are handled in model organism databases (Suzanna Lewis, Judy Blake) (for the curious, all presentations are available from the Jamboree agenda page). After a brief description of the Phenoscape data policy (Todd Vision), Jim Balhoff introduced the curation software tool, Phenex, which Wasila Dahdul followed with a hands-on curation exercise that involved using Phenex and associated ontologies to annotate phenotypes (characters) using EQ syntax.

We then paired up our guest data curators with project personnel and got them started.  In order to really focus them on EQ syntax and ontology development, we gave each of our guest curators pre-curated Phenex files containing taxon lists, character matrices and free text character and state descriptions from several important publications authored by the guest data curators themselves or within their area of specialty.

Inevitably, curators spent a significant portion of the time on ontology development.  Many new anatomical and quality entities, along with their relationships and definitions, needed to be added to either the Teleost Anatomy Ontology (TAO) or the Phenotype Ontology (PATO) before characters could be expressed in EQ syntax. In the process of training curators to submit new entities to the term trackers for these ontologies we recognized the need for small, dedicated ontology development meetings (e.g. focused on one anatomical region such as the lateral line system in fishes, or all geometrical or shape qualities).  The better the ontology, the more efficient curation will be.

The curation workflow itself is significantly different in Phenex from the Phenote tool, which we used in the first Data Jamboree. It were the issues brought up there that guided the development of Phenex, and indeed this time we noticed a significant improvement in curatorial efficiency.  We were very pleased that by the end of the workshop, the curators had curated 156 characters (~312 character states) for approximately 150 taxa, resulting in over 45,000 EQ annotations.

One frequent topic of discussion was the depth of annotation granularity to which characters should be curated. For example, many systematic characters pertain to shape, frequently with complex descriptors. These can be curated to a high level (e.g. fin: shape) or to a more granular level (e.g. fin: anterior margin rounded). Judy Blake, one of our outside advisors, pointed out that there is a continuum between use of a structured vocabulary and free text.  A reasonable guideline is to say that data specific to an individual study may not warrant the extra effort with granular annotation and should be left as free-text while data that can be compared across studies should be annotated with ontologies at a rather detailed level.

One observation was that the higher the level of granularity, the more often post-composition is required. The mechanics in using multiple post-composition windows in Phenex was somewhat confusing to guest curators. Another important topic involved the difficulty of not having a universal standard across systematic studies – e.g. descriptors pertaining to size and shape cannot easily be extended across systematic studies. This same issue came up at our previous data jamboree, too. John Lundberg suggested that for comparisons of size within a study, an internal grading of character states (a numbered scale from 1 for smallest and higher numbers corresponding to larger sizes) could be implemented, and this is something that we plan to try.

Following two days of curation, we conducted an experiment to assess curation consistency among our group of curators, and to identify areas where consistency can be improved, for example by better curator training, upfront ontology development, or user interface enhancements. We wanted to determine how often, and for what reasons, curators choose divergent EQ conceptualizations for the same character and character states. Five curators encoded EQ annotations for the same 10 character and state descriptions, using Phenex as the software tool. Immediately afterward we reviewed the results with the group. Notably, only two of the 10 characters were annotated identically among all curators. The reasons for why the other annotations differed revealed different interpretations of shape descriptors, inexperience and unfamiliarity with the ontologies and software, and lack of adequate terms in the ontologies (e.g. shape) as major hurdles towards consistency between curators. Identifying these hurdles helps us to prioritize efforts on visualization tools, ontology development and workflow and Phenex development.

In parallel to the curation activities, we discussed the Teleost Taxonomy Ontology (TTO), taxon concepts, and intermediate synonyms.  Peter Midford has added the intermediate synonyms from the Catalog of Fishes (CoF), but many additional synonyms are present in the literature that must be added. We determined that the need to associate synonyms to their references/publications will require an OBO request for one or more database identifiers. At first we considered using the CoF publication database to generate dbxrefs rather than hunting for DOIs or generating our own, but CoF doesn’t contain many of our publications.  Thus we will need to maintain references and unique identifiers for them in our own database.

In another parallel activity, pairs of participants met with Jim Balhoff and Cartik Kothari  for demonstrations of the web user interface. Jim and Cartik are creating the database and web-application for the research community to explore and query the data. The central gateways for entering the database that emerged from those consultations are querying by gene, anatomy, taxon, or publication. The use case-level query to be implemented first is to find evolutionary phenotypes that match a mutant ZFIN phenotype, or, going in the reverse direction, to find ZFIN mutants for a set of phenotypes that differ between taxa. We plan to be able to demonstrate the first version of the interface at the educational outreach workshop at the SICB meeting on Jan. 6, where we also plan to conduct a first series of usability tests. (For those curious, there was much more feedback and suggestions we received, which we compiled all on the project wiki). 

Working our way towards that meeting over the coming months, we will put into place the first components of a production-level database and a fleshed out query interface. Aside from the software development efforts, we are planning ontology development workshops in small groups in the coming year, for example for TAO and for PATO, and we tentatively aim our next outreach workshop for ASIH 2009. In the meantime, we will need to review our data curation priorities in light of our high priority use cases and proof-of-concept experiments for queries across ZFIN and Phenoscape.

Curation of evolutionary phenotypes from the systematic evolutionary literature of fishes is central to accomplishing our goal, which is to prototype an ontology-based informatics system to integrate evolutionary, anatomical, developmental, and genetics data. This summer we accomplished a fair chunk of curation, and I’m summarizing this here (see graph below).

During the past year we’ve contacted taxon experts to suggest comparative systematic treatments and to prioritize them based on generality of taxonomic coverage, authors’ enumeration of characters and a published character X taxon matrix. You can see the 76 papers that we ranked as top priority (“A”) and our curation priority in a publicly available Google spreadsheet and on the graph below. We use this document internally to record the nitty-gritty of our progress. (If you see any papers that we’ve missed, please let us know.)

First things first - we needed a OCR formated pdf of each of the 76 “A” papers. It took a summer of searching, interlibrary loaning and downloading docs from university libraries, the BioHeritage Library, and scanning hard copies of electronically unavailable papers. We completed this work for our “A” papers last week.

In order to populate a database with skeletal phenotypes from many species, we need information on the taxonomic distribution of particular phenotypes (i.e., which species show which phenotype). Using Phenote, the software curation tool customized by Jim Balhoff (NESCent) for curation of evolutionary biology literature, we entered taxon lists for each paper by cutting and pasting from the pdfs into a “Publication Taxon” column and into a “Valid Taxon” column. Where published taxon name was no longer valid, it is sometimes designated as a synonym for the valid taxon in the Teleost Taxonomy Ontology, and was thus replaced. Because some intermediate synonyms are not yet in the Teleost Taxonomy Ontology, taxonomic expertise is still required to find valid taxon name for some of the published names. We have taxon lists for 59 of the 76 papers; many of these still require expert review.

Character X taxon matrices make the association between taxa and character states. To save time required for manual entry, we contacted authors directly for their matrices, and some had them available to send to us (thank you!). It’s interesting that about a third of the “A” papers (24/76) do not have published character matrices. We may be able to reconstruct matrices from the text in some cases. We currently have 27 character matrices and 25 remain to be manually entered.

Annotating the skeletal phenotypes in these papers using EQ syntax and ontologies is our current challenge, along with merging old Phenote files into new Phenex format. We are taking a first-pass through the characters, initially coding the easy ones (e.g. Basibranchial 1 one present or absent) and submitting needed terms to the Teleost Anatomy Ontology (to see these terms and comment on them as proposed, please subscribe). We currently have 11 papers curated at this level; our goal is to do a first pass through most if not all of the A papers before our second data jamboree, where taxon experts will review and edit the data.

I want to thank our Deep-Fin sponsored Phenoscape intern this summer, Sonia Crichigno, who worked with our Ontology Curator Dr. Wasila Dahdul to enter taxon lists and character EQs. I particularly want to thank our other summer students Carin Rambow and Leah Mabee, who gathered pdfs, scanned and OCR’d documents, organized and updated the Google spreadsheet, typed in character matrices, and made initial taxon lists. I also want to thank Dr. Miles Coburn, a close colleague of mine and friend of all of us involved in Phenoscape. Miles was an expert on cypriniform fishes, who was reviewing and updating taxon lists and beginning character curation when he was killed in a bicycle accident on August 16th. We all miss him.

As we proposed in an earlier post, we have been developing an alternative to the traditional approach of representing taxonomy in ontologies. This alternative represents species (and currently higher taxa) as individuals in the ontology.

There is another phenoscape ontology that would benefit from the use of individuals: our ontology (ok, it’s really a vocabulary) of research collections of fish. Part of our process for curating anatomy papers involves constructing a list of all the specimens reported in the paper (generally the author includes this in the paper, but we enter it to facilitate annotation). The specimen lists consist of collection names and numbers. Although research collections are supposed to have standardize names and abbreviations, practice does not always reflect these standards: the same collection may be abbreviated in different ways by different authors, and collections occasionally merge and the smaller collections might disappear into larger ones or be renamed.

So we constructed a vocabulary of fish collections with their 4-5 letter abbreviations, longer name and possible synonyms. Now a research collection contains individual specimens, but these are parts of a collection, not subtypes of a particular collection, and therefore, the collection is best represented as an individual, not a class. So, the collections vocabulary, represented as an OBO ontology, should ideally use OBO instances (which exist to represent individuals), rather than terms for the collections.

Unfortunately, the current version of OBO-Edit, although it will read and save ontologies containing Individuals, provides no facility for either viewing or editing these individuals. Furthermore, although there are several OWL editors that would allow us to work with Individuals, there are no OBO<->OWL translators that understand what to do with OWL individuals, so they are just omitted in the OBO translation, even though the OBO file format, since version 1.2, has supported Instance ’stanzas.’

Several of us recently discussed this issue with Chris Mungall, who mentioned a similar issue with the GAZ ontology of geographical terms, though the developers of GAZ are not as concerned as we have been with the Individual/Instance vs. Class/Term distinction. Since adding support for Individuals would impact the final release of OBO-Edit 2.0, which we all want, and as an OBO-Edit plugin was considered too risky (e.g., would likely introduce difficult to detect and fix bugs), Chris proposed an ‘out board’ solution.

This solution would involve adding an instance_of relation to any ontology that wanted to use individuals. Although instance_of is not one of the ten relations listed in the published OBO relations ontology, it is defined as a primitive and used throughout Smith et al. (2005).  Entities that were intended to be individuals would be defined as terms, but rather than link to their parents though is_a relations, they would use the locally defined instance_of. The out board tool will be able to convert OBO stanzas that linked to parents through instance_of back and forth between term stanzas and instance stanzas. With term stanzas, individuals would be syntactically classes, but visible to OBO-Edit, whereas when they were instance stanzas, they would be correct, but invisible to current versions of OBO-Edit. Eventually, OBO-Edit and its reasoner(s) will include support for individuals.

This seems like a workable solution and the best we can expect given current limits of time and people.

Our first educational and outreach event “Evolutionary Biology & Ontologies Workshop” was held at the Evolution meetings in Minneapolis, Minnesota (June 20, 2008), and we felt it was a big success.  We had lots of enthusiasm and over 50 attendees for this all day workshop, which was organized by the Phenoscape PIs (Paula Mabee, Todd Vision, Monte Westerfield), NESCent and Barry Smith from the National Center for Biomedical Ontologies (NCBO). 

We need to especially thank all our speakers for their excellent presentations, They not only gave the audience a varied introduction to ontologies, but also a set of examples of the integrative questions that can be answered using them.   The slideshows for all of these talks are on our wiki but I thought I would provide a brief overview of each one right below as a summary of the workshop. The use of ontologies is just emerging in evolutionary biology, and it is an exciting time to be involved in this field.  As we move forward to use ontologies in evolutionary biology, we discover new requirements and challenges — for example, the challenge of how we create ontologies that are interoperable — so that we can ask big questions that span not only taxonomic groups (such as bees and fishes and mouse and fly) but different knowledge domains (such as phenotype, evolution, and genetics, genomics, medicine).

So here is an overview of each talk.

Barry Smith (NCBO and Univ. Buffalo) kicked off the workshop with “An Introduction to Ontology for Evolutionary Biology“.  He talked about the GO in relation to “doing biology across the genome”, and he described the domain coverage of the ontologies in the OBO foundry.  He introduced basic ontological concepts and relations, including several new ones that must be made in creating ontology resources for evolutionary biology.  He is currently working with a number of groups including EoL to develop an Environment Ontology (EnvO).

Chris Mungall (Lawrence Berkeley Labs) presented “PATO: The Phenotype and Trait Ontology“.  He focused on why we need ontologies (for example, to solve the data integration problem), introduced “phenotype” and the newly proposed Homologous_to relation. 

Monte Westerfield (Univ. Oregon, Zebrafish Information Network) talked about the use of ontologies in “Linking Animal Models and Human Diseases“.  Just as BLAST can be used to connect animal genes to human genes, shared ontologies and syntax can connect mutant phenotypes to candidate human disease genes. He talked about the importance of curation consistency and how it can be measured and increased.  His proof of principle example, comparing ontology-annotated phenotypes in human and zebrafish demonstrated that genes associated with human diseases can be identified.

It was then my turn to introduce Phenoscape in my talk “An Introduction to the Use of Ontologies in Linking Evolutionary Phenotypes to Genetics“.  Just as animal model phenotypes can be compared to human phenotypes using ontologies, so can animal model phenotypes be compared to phenotypes of multiple species - thus linking comparative evolutionary data to model organism genetics.  I presented several use cases for fishes and described the rapid progress that we have made in developing new ontologies, extending existing ones as well as some of the challenges involved in multispecies ontologies.

Jim Balhoff (NESCent, Phenoscape) presented the software that he has developed for curation of evolutionary phenotypes “Phenote: Curation Software for Evolutionary Phenotypes“.  Originally developed for ontology-based annotation of genetic mutant phenotypes and widely used by model organism curators, Jim described how he has extended and reconfigured Phenote to handle comparative evolutionary data, and he also spoke about his plans to integrate ontology-based phenotype descriptions into the NeXML file format.

Melissa Haendel (Zebrafish Information Network) presented “The Common Anatomy Reference Ontology (CARO) and queries across species“.  She described current anatomy ontologies and why we build them, and the standardization and integration of anatomy ontologies using CARO.  She explained how the issues created by homology can be addressed, including the draft definition of the Homologous_to relationship, and she described how homology data can be used with ontologies to query across species. 

After lunch we had about a ½ hour of “Lightning Talks” during which workshop attendees had an opportunity to introduce themselves and give brief (2-3 minutes) descriptions of their interest in ontologies.  Several attendees seized the chance and made the audience aware of their ontology building efforts.

After that, Ann Maglia (Missouri Univ. of Science and Technology, AmphibAnat) resumed the regular talks and spoke about her group’s work in the “Development of the Amphibian Anatomical Ontology“.  This ontology was driven by the need of the amphibian community to standardize their terminology and thus integrate their comparative work.  She is exploring automated techniques to reduce manual efforts and enhance existing, manually created ontologies.  Ann also described the relational database (RDBOM), the modularization, and web-based community curation methods built by her group and community. 

Martin Ramirez (Museo Argentino de Ciencias Naturales) in his talk “Ontologies, Image Databases, and Evolutionary Biology” described the use of the spider ontology to link images from the spider Tree of Life project to a phylogenetic dataset and a body of image annotation.  He discussed a series of challenges, such as issues involving the alignment of ontologies using homology relations and the “combinability” of ontologies.

Todd Vision (Univ. North Carolina, NESCent), presented “Ontologies and the Identification of Candidate Genes for Complex Traits“.  He considered the qualities of “good” candidate genes and how they are typically discovered and ranked, and then described a test case where ontology terms were used in picking candidate genes computationally (using an approach named CAESAR) for Type 2 Diabetes.  This approach has a promising application for evolutionary traits in non-model organisms.

Peter Midford (Univ. Kansas, Phenoscape) introduced taxonomy ontologies with his talk “Names, Ranks, Clades, and Taxonomy Ontologies” — what’s in them, how they are built and why one would use one.  The issue of what species are ontologically, more specifically whether species are classes or instances or both, impacts not only the way an ontology is built but also what kinds of assertions can be inferred or not.

Suzanna Lewis (Lawrence Berkeley Lab, BBOP) and I wrapped up by listing some of the challenges for the evolutionary biology community that became apparent over the day. 

Especially prominent was the need for communication among evolutionary biologists developing ontologies so that they can be developed from the outset to be interoperable.  This and many related comments from attendees suggested that an NSF-RCN (Research Coordination Network) proposal might be the best next step and an effective mechanism to fund and promote communication among and between evolutionary biologists, ontologists, and model organism biologists.

As a result, we are now beginning to gather ideas, objectives, and possible participants for an RCN proposal. If you have thoughts on this, or if you are interested in participating, we would love to hear from you - leave a comment here, or email me (pmabee{at}usd{dot}edu).

Peter Midford and I recently attend the two-day Relations Ontology workshop in Denver, Colorado. The goal of the meeting was to further develop the Relations Ontology (RO) by moving relations from RO proposed (ontology for relations yet to be officially added to RO) to RO, and adding any new relations proposed by attendees. Impressively, almost all of the items on the original agenda were covered, and an ‘action list’ was produced to focus efforts subsequent to the meeting.

Phenoscape had several relations on the agenda, and Peter and I have summarized the discussion, definitions, and examples used.

On Monday afternoon, Wasila and Peter presented proposals for the following relations for the section organized by Suzi Lewis and Melissa Haendel on spatial relations:

1. Overlaps

Peter presented examples and definition for overlaps. Overlaps is particularly useful for Phenoscape so that we can represent a joint without having to create new terms for the parts of bones that participate in the joint. We can simply say, for example, that bone X overlaps joint XY. After brief discussion, it was decided that this relation will be added to RO with the proposed definition.

Examples:

bone overlaps joint

axon overlaps axon tract

definition:

X overlaps Y =def. for every t and every x, if x instance_of X at t, then there is some instance y of Y at t such that (x overlaps y at t)

where

x overlaps y at t =def there is some z such that z is part_of x at t and z part_of y at t

Note that it can be the case that X overlaps Y as thus defined, even though Y does not overlaps

2. Attached_to and connected_to:

Wasila presented examples of these two relations and definitions given by the FMA (although the FMA definitions are reversed from how we use them in our use cases). There was discussion of whether or not it is useful to distinguish between fiat and bona fide boundaries, whether a single “connected” relation could suffice for all use cases, and whether a force-connected definition would be better to use. However, the overriding point was that biologists clearly make the distinction between attached and connected, and it was decided that attached_to and connected_to would be added to RO pending development of their formal definitions.

Examples:

tooth attached_to bone

muscle attached_to bone

axon connected_to neuropil

3. Spatial Ontology

Wasila presented examples of how Phenoscape uses the Spatial Ontology, and much discussion centered on how to improve this relatively new and little used ontology. Some classes (e.g., anatomical region, anatomical surface) should be moved to CARO, and reference images would be helpful in providing examples for the different anatomical axes.

Homologous_to relation

We presented a proposal for homology relations on Tuesday morning. Melissa started with a couple of background slides and Peter continued with a presentation of the definition we developed as a group on Monday night. That definition was revised again on Tuesday afternoon and the Tuesday afternoon revision can be found below. It, along with earlier revisions is available on the wiki page (http://www.bioontology.org/wiki/index.php/RO:Main_Page#Proposed_homologous_to_relation).

Directly_descends_from - This is currently defined as follows

Example: Child’s arm directly_descends_from parent’s arm

definition:

x1 directly_descends_from x2 iff there are y1, y2 such that:

• y1 is an organism
• x1 is an anatomical structure
• x1 part_of y1
• y2 is an organism
• x2 is an anatomical structure
• x2 part_of y2
• y2 is a parent of y1
• the genetic sequence that determined the morphology of x1 is partially a copy of the genetic sequence that determined the morphology of x2*.

* This clause still needs some work. Jim Balhoff added the following critique in discussion on the Phenoscape project list:

[Something that jumps out at me in the definition of directly_descends_from:

I would not say that genetic sequences "determine" any morphology. I would prefer something like "participates in the development of" the morphology of x1. Anyway, I don't see genetic sequences as an absolutely necessary component of homology (although they would very often be an important component).]

Descends_from - This is currently defined as follows:

Example: My arm descends_from ancestral primate forelimb.

definition:

descends_from is the instance level relation which is the transitive closure over directly_descends_from:

A in B descends_from C in D:

For all A(a) -> exists b, d, c: B(b) & C(c) & D(d)

• a part_of b
• a descends_from c
• c part_of d

(Note – B must be a subclade of the clade descending* from D)

Homologous_to - Defined for classes of anatomical structures and clades

Example: Human arm homologous to robin wing.

definition:

A1 in B1 homologous_to A2 in B2 iff exists A3, B3:

• A1 in B1 descends_from A3 in B3 &
• A2 in B2 descends_from A3 in B3

(Note B1 and B2 must both be subclades of the clade descending* from D)

(*In the genealogical sense)

After presenting the definition, Peter presented a number of examples of character optimization – these included a schematic overview and a tree of bilateralians with a reconstruction that indicated that a heart was not present in the most recent common ancestor of bilateralians. There were also other examples that indicated different common ancestors and presence/absence of homology for traditional and Ecdysozoa/Lophotrochozoa-based phylogenies. The presentation finished with a discussion of the homology evidence codes and their use in Phenote.

We are uncertain of the need to include time (e.g., replace the atemporal usage of exists with temporally scoped exists and existed). Fabian suggested that just to be on the safe side we can include time – it’s not useful but it could stop some people complaining and it won’t affect the logic at all.

This definition mentions clades, avoiding the issue of whether taxa are individuals, classes or both.

We will be revisiting these definitions in the future.

The Teleost Anatomy Ontology (TAO) is a multi-species anatomy ontology for teleost fishes.  In this first post about the TAO, I’ll introduce the structure of the ontology and its development, and discuss some of the challenges we’ve come across in building a multi-species ontology. You can browse the TAO by using the NCBO BioPortal.

Development of the TAO

The TAO was originally cloned on the Zebrafish Anatomy Ontology (ZFA), and the upper level classes in both ontologies (e.g., anatomical structure, portion of organism substance) are based on the Common Anatomy Reference Ontology (CARO).  Current development of the TAO is concentrated on the skeletal system, which is often the focus of morphology-based evolutionary studies in ichthyology, owing to the fact that it varies significantly across teleosts and is well-preserved in fossil specimens.  Terms (classes) in the TAO have relationships to other terms, for example:

extrascapular is_a dermal bone

extrascapular part_of pectoral girdle

Growth of the TAO is enabled by term requests from data curators and the ichthyological community, and the TAO is updated frequently to include new terms, refined definitions, synonym additions, and structural changes. We currently maintain manual synchronization of the TAO and ZFA, but a synchronization tool is in development and will help us automate this process. Synchronization between the multi-species TAO and a single-species ZFA presents the challenge of maintaining equivalent structural hierarchies between the two ontologies.  This is because the addition of new terms to the TAO can sometimes require the creation of parent terms necessary for the TAO but redundant for the ZFA (see TAO and ZFA synchronization). 

Synonyms

Accommodating different names (synonyms) used to refer to the same bone is another challenge in building a multi-species anatomy ontology.  Ontologies accommodate synonyms by designating a primary term name along with assigned synonyms; for example, extrascapular is a primary term in the TAO, with supratemporal recorded as its synonym. Of course, different communities of taxonomic experts may have preferences for one term over another, and we plan to allow users of our future database to choose their ‘preferred term’ when using our tools. 

Representing synonyms is usually a straightforward exercise, but we’ve recently come across a situation in which a bone name will need to be designated as both a primary term name and a synonym of another primary term. The infraorbital series is a set of bones that typically lie below and posterior to the eye in fishes.  The general condition for teleosts is that the last bone of the series is the dermosphenotic, which is often times referred to in the ichthyological literature by its position in the series.  For example, the dermosphenotic is called infraorbital 5 in cyprinid fishes (including Danio rerio), whereas it is called infraorbital 6 in characiform fishes. In order to represent the usage of these terms in the TAO, dermosphenotic will be a primary term with several synonyms including ‘infraorbital 5’ and ‘infraorbital 6’, names that are also primary term names. I’m currently working up definitions for these terms, and you’ll see the proposed definitions shortly on the TAO term tracker.

New anatomical relations

The addition of new terms is not the only way that the TAO is developing.  Peter Midford and I attended the Relations Ontology meeting earlier this week to discuss our need for several new anatomical relations, including attached_to, connected_to, and homologous_to. In an upcoming post, we’ll summarize the results of the meeting and provide examples of our use of the new relations.   The new relations will be added to the next version of the Relations Ontology, and therefore available for use in the TAO.

Searching as well as visualizing the Teleost Anatomy Ontology (TAO) on the NCBO Bioportal was broken for more than a week but has been fixed since Friday.

The other good update from the Bioportal development is that terms can now be found by their synonyms as well. For example, try searching TAO for the ‘dermosphenotic’, which at present isn’t the name of a term in the ontology. Instead, you get the ‘infraorbital 5‘, for which dermosphenotic is a synonym.

Fish morphologists will note that this is actually problematic, since in reality the dermosphenotic is the synonym for the last infraorbital bone. In zebrafish, with which we seeded the TAO, this is indeed the 5th in the series of infraorbitals, but in other clades of teleosts it is the 6th or yet another one. But that’s another story, which we’ll highlight in a forthcoming post on building the TAO.

We have created a new mailing list, obo-taxonomy, under the OBO (Open Biomedical Ontologies) umbrella. Our motivation for this new forum is to really open up the discussion surrounding the issues of what should be a proper ontological representation of taxonomy and phylogeny, for example proper semantics of the relationship between taxonomic groups, and between specimens and species. If you care about or have thoughts or opinions on these and related questions, we encourage you to subscribe to this new list.

One of the key components in our efforts to code the evolutionary diversity of phenotypes in a semantically rich and explicit manner is to have species or higher-level taxa be terms from a logically consistent ontology. Eventually we will want to reason over taxonomy ontologies and entity-quality statements to infer the evolutionary change of phenotypes between different clades of fishes, and hence the semantics of a particular ontological representation of a taxonomy or phylogeny will have bearing on the kind of inferences we will be able to make. It will probably come as no surprise that we have since found ourselves discussing the issues and what they might mean for us in quite some depth (see for example Peter’s previous post on how we built the first version of the TTO, and our work in progress on the Taxonomic Rank Ontology).

It turns out that questions of using a hierarchical controlled vocabulary or ontology to denote the taxonomic species is coming up on an increasing basis in the contexts of annotating biological experiments, observations, comparative analyses, and results, and probably elsewhere too. In fact, creating this group was instigated by a conversation about the NCBI taxon ontology within the OBI (Ontology of Biomedical Investigations) Biomaterials group.

As an aside, for those of you who dislike the SourceForge.Net mailing list interface, the list is mirrored as a Google Group, and you are welcome to use the Google Group interface to subscribe or post. 

One of the two main ontologies developed and used by the Phenoscape project is the Teleost Taxonomy Ontology (TTO). Although the Phenoscape project is focused on the Ostariophysi, the TTO covers not just teleosts, but all the species listed in Bill Eschmeyer’s Catalog of Fish. This post will discuss how the current TTO was constructed and the work flow we use to update it. A later posting will discuss the effort to update the ontology to better represent current thinking about metaphysical status of species and other taxonomic terms.

What’s in the TTO?

The TTO contains slightly over 35,000 terms, of which 30,385 are species, 5045 are genera, and 542 are families. The ontology is organized as a traditional taxonomy, which is also refered to a class hierarchy - thus the species Danio rerio is a subclass of the genus Danio, which is a subclass of the family Cyprinidae. In addition to terms from the Catalog of Fish (CoF), we have added a small number of ‘higher level’ taxa, which allow the TTO to have Craniata as its root.

The TTO also contains a set of taxonomic rank terms - ‘genus’, ‘family’, ‘order’, etc. Most of the taxa named in the ontology are tagged with one of these rank terms, using a property called ‘has_rank.’ This structure is virtually identical to the NCBI taxonomy ontology that is periodically generated by Chris Mungall. The current structure is diagrammed in the figure below. I have constructed a separate ontology of rank terms, which will be submitted to the OBO community in the near future.

The TTO also contains over 38,000 taxonomic synonyms, mostly at the species level, but there are also some at the level of genera.

We have also consulted with taxonomic experts for several Ostariophysian families, including Siluriforms, Characiforms, and Cypriniforms. One benefit of the contributions from area experts is the addition of several fossil taxa, particularly for Siluriforms. Although fossil taxa are not included in the CoF or in the NCBI taxonomy, which focuses on taxa which have molecular data, inclusion of fossil taxa is consistent with the use of this ontology to support annotation of morphological characters. For taxa that have not entered the TTO as an entry in the CoF, we try to include a doi or similar identifier as a database cross reference.

For taxa that are mentioned, but are not actually described in a publication (e.g. Abramites sp. appearing in Fink and Fink (1981)), we include the source publication as part of the name, for example: “Abramites sp. (Fink and Fink 1981).”

There is a Wiki page that lists all the changes we have made relative to the last version directly generated from the CoF. This is currently from November 2007, though we will be generating an update from the January 2008 CoF in the near future.

Constructing the TTO

The TTO was constructed from a database dump of the Catalog of Fish that was very generously made available to us by Bill Eschmeyer and Stan Blum. The dump consisted of three tables: species, genera, and lineages. Each table consisted of rows of names, with the status of each name and, if appropriate, whether the name was currently valid or a synonym. Each row also indicated the higher level group that subsumes each term. I constructed a tool ‘TTOUpdate’ to build the OBO format ontology from these three tables.

The tool is currently capable of doing limited updates from tables in a special format. Otherwise, updates are currently done by hand, though the next version of the TTOUpdate tool should be more flexible. The next version will also support generation of ‘intermediate synonyms’ by extracting taxon names from the history comments. Currently only names that appear as either current names (which might however, be marked as synonyms) or original names will be extracted from the CoF into the TTO (as either valid term names or synonyms).

Browsing the TTO

We do not have a tool specifically for browsing the TTO, however, there are several options for doing so. The TTO is available on NCBO’s bioportal, though we have been having some difficultly with that site. However, the TTO is available as a text file (OBO format), can be visualized as a tree, and can be searched from here.

If you are familiar with OBOEdit, you can download the text version from any of the usual OBO respositories (e.g., the source forge cvs ). The cvs browser will also let you see another view of the changes to the TTO, in terms of tracker items and other requests that each change addresses (I do try to write meaningful commit comments).

Aside from the Evolutionary Biology and Ontologies Workshop that we organize together with the NCBO at the Evolution Meeting 2008 in Minneapolis, we will also be presenting a poster. For your advance pleasure, here is the abstract that we wrote, only to realize that registration for a poster presentation actually did not require or even offer any abstract submission.

Title: Linking evolutionary morphology to genomics using ontologies

Authors: Paula Mabee, Jim Balhoff, Wasila Dahdul, Hilmar Lapp, John Lundberg, Peter Midford, Todd Vision, Monte Westerfield

Although systematists have collected comparative morphological data for decades, it has been rendered in free-form text and is not computable. Even seemingly straightforward comparisons across taxa, such as examining evolutionary patterns of characters from the same body region (e.g. head) or of the same quality (e.g. shape changes), cannot be made broadly without a database and some means of computation. Ontologies, or structured vocabularies with semantically defined relationships, provide such means and facilitate integration with other databases, including genetics databases from model organisms. Using ostariophysan fishes, we are prototyping a system with an ontology-based character database and generalizable tools to integrate evolutionary and model organism phenotype data. This system is designed to meet the top-priority needs of the evolutionary community concerning development and evolution of morphology. These include mapping characters on trees and identification of clades in which traits vary (quick retrieval of character state distributions), correlation matrices of traits, identification of candidate genes, and phenotypic BLAST to discover similar phenotypes and similar OTUs. Such queries will help address large-scale questions about the genetic and developmental regulation of evolutionary morphological transitions.

To date we have laid the groundwork for character annotation through the development of multiple ontologies, curatorial tools, and careful consideration of how to deal with homology, types of evidence, and voucher specimens. We have developed a multi-species anatomy ontology for teleost fishes (Teleost Anatomy Ontology) using the existing zebrafish anatomy ontology as its core. [The TAO is being expanded and refined by the broader ichthyological community]. We have developed new components for Phenote, an open source software application, to facilitate the annotation of evolutionary characters from the literature using terms from the teleost anatomy ontology in combination with terms from the Phenotype and Trait Ontology (PATO) and Spatial Ontology (BSPO), following and adapting the Entity/Quality syntax of model organism databases. New components of Phenote include a phylogeny view for graphically selecting clades to annotate, a taxon list manager for entry of voucher specimen identifiers, a data filtering interface, and more. We are documenting character observations from individual species whose names are drawn from a taxonomy ontology (Teleost Taxonomy Ontology) that we developed based on an expert database, the Catalog of Fishes. Phenoscape will enable researchers to investigate relationships between evolutionary changes in morphology and the phenotypes seen in genetically characterized developmental mutants of zebrafish.

We finally set up a Phenoscape project on SourceForge.net. At this moment there isn’t much there in the code repository, but within the next days we’ll deposit some scripts that we had to develop for OBO-format ontology generation (for example, for initiating the TTO from the Catalog of Fishes) and for massaging the tracker messages auto-generated by the OBO term request trackers to be more user-friendly. The source code repository will also be the home for some controlled vocabularies that (only?) we need for our project (such as the one for museum codes).

You might wonder why this has taken almost a year since we launched Phenoscape in June 2007.  The reason is that until just recently our software as well as ontology development activities have built on existing projects (such as Phenote on the software side or PATO on the ontology side) that already had their community infrastructure, or given rise to new community ontologies (such as the TAO and TTO) that each received their infrastructure within the OBO umbrella.

However, as we start shifting our focus this summer to developing the database and web-based user interface for the EQ data our curators will soon be generating in increasing amounts, our own code repository will be the natural home for much of the code we’ll develop.

The first Phenoscape Data Jamboree (we are scheduled to have one each year) is starting today at NESCent. The event brings three fish morphologists external to the project (Miles Coburn, Kevin Conway, Mário de Pinna) together with our morphologist, ontology, and informatics personnel. In addition, Nicole Washington (NCBO) and Martin Ringwald (Jackson Laboratory), two experts from communities that have made considerable strides in bringing ontology-driven and semantically explicit approaches to bear on annotating gene function and mouse phenotypes and gene expression, respectively, are here to serve in an advisory role.

We have three overall goals with the Data Jamboree. The first is to train fish morphologists outside of our project to understand how to use the tools (primarily Phenote) and ontologies to generate phenotype annotations compliant with the EQ approach. 

The second is to develop, review and where necessary refine best practices guidelines for annotation, to ensure that different morphologist generate the same, or at least semantically equivalent phenotype assertions from the same character descriptions.

The third goal is to generate actual data. Hence, after an initial training exercise, over the next two and a half days the morphologists and our curators will be going over a pre-selected set of publications and translate the character descriptions into EQ-compliant assertions, and the informaticists among us as well as our advisors will look over their shoulders to see what needs refinement.

Released new Phenote version (download). It has the following new features:

• Changed version scheme again, hopefully for the last time
• Bug fixes to in-table editing
• Added TAO to the Quality field for use in post-composition
• Included built-in Phenoscape interface layout
• Re-ordered input fields to better match curation workflow