2Aug06teleconf

From Phyloinformatics
Jump to: navigation, search

2 Aug 2006 teleconference

Summary notes from Arlin Stoltzfus

We began at about 2:40 EDT with some introductions-- Hilmar Lapp, Aaron Mackey, Todd Vision, Mark Holder and I were on line. Todd explained NESCent's interests, the availability of support to host meetings, and the desire to identify objectives for a proposed hackathon in November. Mark provided information about CIPRES, its goals and its work so far, and throughout the discussion he continued to supplement this. Hilmar and Aaron mentioned their experience as BioPerl developers. Hilmar has been with NESCent for several months.

As an initial attempt to reach broad consensus, I asked if we were all comfortable with the idea of focusing the hack-a-thon on providing support for CORBA-brokered CIPRES services (the idea that Todd suggested to some of us about 2 months ago).

Unexpectedly (for me) it turned out that this issue dominated the rest of the discussion. Here are some of the points that came up along the way.

Aaron asked if there would be public access to the server. Mark said that CIPRES would not be providing compute power, because the demand just from the TOL community would "bring the server to its knees".

Hilmar asked about what kind of services would be provided. Mark provided some more detail about the CIPRES services, which would include alignment, phylogeny inference, and supertrees.

Arlin pointed out that, if CIPRES/CORBA design turns out to be useful, then it can be replicated on other servers and expanded to other services, by analogy with bioinformatics servers that are set up and maintained by individual departments or institutions to provide services to their users.

Hilmar pointed out systems such as PISE and EMBOSS that provide interfaces to a variety of tools (Arlin notes that PISE seems to be a real success story-- I have attached a PDF about it). Aaron points out that these systems require a large initial investment of time and require expert maintenance, but the server can then can be useful to many.

Mark noted that the learning curve for the CIPRES would be steep and that the interface currently was not well documented. Arlin suggested that expanding the user base for evolutionary analysis tools, and supporting high-throughput analyses would seem to depend on having lightweight standards and course-grained interfaces.

There was some discussion about graphical vs. command-driven interfaces. Mark indicated (I think) that both are possible. However, the command-driven interface is not well documented.

There was a small discussion of workflow software that mentioned Kepler and TAVERNA. (Arlin can provide references to these for those who are interested).

We discussed some possible demonstration projects that would involve phylogenetic analysis or hypothesis-testing with some large set of sequence families. Aaron raised the question of whether there was anything that the CIPRES server could do that could not be done with BioPerl. Mark suggests perhaps not.

A discussion involving several people (Aaron, Hilmar, Arlin) raised the issue that BioPerl has many holes in it, i.e., it defines "Bio::Run" interfaces to various tools but the interfaces actually have not been written yet. Aaron suggested that a possible project would be to fill in these holes in BioPerl's evolutionary analysis capabilities.

Arlin mentioned his need to finish up at 3:30 EDT due to other commitments. Hilmar suggested helpfully that perhaps each of us could offer our ideas for possible objectives, then Todd could sort through these and report back to us, presumably after discussing the issue within NESCent. There seemed to be a consensus on this course of action.

CONCLUSION AND ACTION ITEMS

The ultimate outcome was that concerns were raised about how large a community would benefit ultimately from software development focused on the CIPRES services. As a result of these concerns, we did not reach a consensus on specific objectives for a November hack-a-thon. However, there was a consensus on Hilmar's suggestion to propose objectives.

ACTION ITEM: propose specific software development objectives for a hack-a-thon in November and relay these to Todd.