IET BioSysBio 2009

Frank and Dan have already blogged about this year’s BioSysBio conference in Cambridge (23rd-25th March). I just thought I’d add my thoughts to theirs.

I don’t get to go to many conferences. The nature of my work doesn’t really demand it, but about once a year it does me good to reconnect with some cutting edge science, and get a good idea of developments in the field as a whole.

Before now, ISMB has been the conference of choice, as the largest gathering of bioinformatics types, it certainly was the obvious one. But in recent years it has become a cumbersome beast. Multi-tracked and vast, hard to pin down stuff you want to hear, often disappointing when you do find something. So this year we cast about for something smaller and fresher. We had heard good things about BioSysBio last year, and it certainly looked promising, so we made our decision.

And boy, was it the right decision. Small enough to be single track, there were very few choices to make in terms of what talks to attend (actually there were none, there was only really one parallel session, workshops on the Tuesday afternoon, and I was obliged to be at the ONDEX one, since I was helping out). This meant that instead of skipping between halls, missing bits and pieces of talks, and sometimes not bothering at all, I sat in one place, pretty much for 3 days straight, and listened to everything.

Highlights were the ethics and biosecurity debate, with a fabulously engaging talk from Drew Endy; showcases of the importance of transcription initiation and elongation from Marko Djordjevic and Andre Riberio; an excellent Synthetic Biology talk from a man apparently inspired by the iGEM competition, Philip LoCascio; and a couple of excellent videos of lab robots hard at work (Adam the Robot Scientist, and another in the final paper talk of the conference by T Ben Yehezkel).

Wordle of #biosysbio tweets

Wordle of #biosysbio tweets

Next year I would happily micro-blog the conference again. This was my first conference since I joined Twitter and FriendFeed, and I was unsure about how I (and my followers) would feel about really going hard at the live updating of the conference experience. I think, though, that those of us who Tweeted provided an idea of the content being presented to those who could not attend, and the feeling I got from the feedback we received, and the fact that not a single person unfollowed me in the three days, is that we were providing a useful service. It has also provided me with a useful resource, a set of notes on the event produced by a crowd, not just me. Search for #biosysbio to see what I mean. Oh, and no review of this conference would be complete without a mention of Ally’s blogging, in which she chronicled pretty much every single talk, except her own (I did that one!)

I do think that for future events I would create threads on FriendFeed for each talk, and group my thoughts about it there, then tweet the URL of the FriendFeed post – this might make things a little less noisy.

Coming back from a conference feeling exactly how you should feel, refreshed, invigorated and excited to get on with your own work, is a great thing. For this feeling alone I will be returning to BioSysBio next year.

Saint: A lightweight SBML annotation integration environment

Allyson Lister

CISBAN, Newcastle University

This post is an homage to Ally’s own herculean note taking style, since she can’t blog her own talk.

Saint has been developed to help modellers get information into their SBML models really quickly. Ally shows a picture of a model describing neuromuscular junctions (standard biomodel). This model contains terms which are descriptions, and the mathematical model. The maths doesn’t know anything about the underlying biology. For example, actin is just a label, there is no implicit knowledge contained in that label (ie actin is a protein, invoved in the cytoskeleton etc).

Short intro to SBML: SBML is a standard format, which is widely used. it stores the maths and enables linking to the underlying biology.

So what do we know about actin –

  • its a protein (UniProt)
  • interactions? (Pathway Commons, STRING)
  • reactions and parameters (SABIO-RK, BRENDA, KEGG)
  • vocab (SBO, GO)

Now we can use the MIRIAM standard to annotate the model with the above information.

When building a model, you need to add info to things like species, name, reaction, compartment
Annotation and SBO term sit between the model and the biology information – these can be used to retrieve the information from the databases. This has to be done manually currently, this is hard and is often not done exhausively, or even at all.

Saint enables automation of this procedure. It already links to a number of data sources – MIRIAM, UniProt, STRING, SBO, Pathway Commons. Reduces effort on the part of the modeller. Saint is lightweight and easy-to-use. Useful as a first pass annotation tool, or to add annotation to an existing model.

How Saint works:

  • import SBML into Saint
  • Saint then searches for appropriate annotation
  • and presents this annotation, and allows to to accept or reject the changes

Ally is using a model produced by Carole Proctor in CISBAN as an example run-through of Saint.

Saint does some validation via libsbml on import of an SBML model. The tool then presents a list of species found in the model, these can be hidden if you don’t want to retrieve information on them. Zoom into ‘Ctelo’ for an example – a plus next to the name of the species shows the annotation already available in the SBML model (‘known’ information). So we can se that Ctelo is a Capped Telomere. You can decide which species you want to annotate, and which datasources you want to retrieve that annotation from.

Queries are made from datasources by a Master Asynchronous Query Service – once information becomes available, it is immediately visible in the UI (as an ‘inferred’ tab), and you see a ‘New Annotation found’ message. Once Saint has retrieved annotation, the user can choose which annotation he wants to keep, and how this information links to the species in the model (is, part etc – MIRIAM terms)

CDC13 = polypeptide chain, nuclear telomere cap complex, protein binding, single-stranded telomeric DNA binding, telomerase inhibitor activity.

Future work – more data sources, use of species type, better support for non-systematic names, adding software source attribution, incorporation of SBGN (Systems Biology Graphical Notation) for better display.

Personal comments – good job Ally – hope I did it justice!

Ally’s standard disclaimer:

Please note that this post is merely my notes on the presentation. They are not guaranteed to be correct, and unless explicitly stated are not my opinions. They do not reflect the opinions of my employers. Any errors you can happily assume to be mine and no-one else’s. I’m happy to correct any errors you may spot – just let me know!