I am trying to get started using BioMoby, well, actually trying to determine if it is right for me. I have a set of annotations that I have generated using some sequence data and some filters. I want to do some manual validation of these annotations (actually I want to work with the underlying sequences) using some BioMoby services. I am wondering if BioMoby allows any arbitrary sequence to be an input for a workflow in, say, GBrowse Moby, or if I am limited to the annotations currently in the Moby namespace. My annotations are already in a GBrowse, but I am just wondering if the sequences that go along with these can act as inputs to a workflow. All examples I have seen start with a GenBank or some other publicly available annotation. Because my annotations are not in a public database, I would want to start with sequences. Any BioMoby experts out there want to talk me through this?

Thanks,

Ryan

Hi Ryan,

The first thing is to clarify what BioMOBY is. Then we can get going on the workflow you need.

BioMOBY is an ontology-based registry system for web services in bioinformatics. Its main point is to make it easy to build distributed workflows. In a nutshell, a service provider (say the owner of a sequence database, a micro-array database, or an analytical tool like blast), says "hey, I want to provide access to my code/data via web services". They build their service, define the XML inputs and outputs using objects from the moby datatype ontology (which they can add to if needed), and put a description of the service up in the moby service registry. The delineation of exactly what syntax goes in and comes out makes it possible to programmatically discover chains of services (workflows) automatically.

make sense so far?

Yes, so far so good.

So my specific question at this point is as follows:
Can arbitrarily defined annotations of a certain genome (say, human) that are not defined in a registered namespace still be used as inputs in such a workflow?

Thanks for your fast response,

Ryan

The answer is that anything represented and registered as a moby object can be used as the input or the output to a biomoby service and thus can be incorporated into a moby workflow.

The BioMoby registry doesn't 'know' anything about biology. Any knowledge needed (like what an annotation or a genome or genome coordinate is) would have to be represented in a moby object or in a service description. So, the question is how to represent your genome annotations as moby objects. How do you have these annotations represented right now? Do you have sets of actual sequence files with some sort of descriptive information? Fasta files? Genome coordinates???

You don't need a namespace if one doesn't exist - its just for identification of your entities. The system can work with only declared objects types (e.g. there are services that take in any DNASequence and output AnnotatedDNASequence or BlastResult).

You might have a look at some of the other (non-gbrowse) moby clients - like
remora or the others listed here. (Note that mine (Ahab) doe not seem to be running at the moment).

Sorry - little rambling this morning .