SSSB Liveblog: Saturday Morning

Well, the big day has arrived and Jason and I are sitting with 40 great synthetic biologists (some young, some old, all innovative) at the Standards in Systems and Synthetic Biology 1 Workshop.

9:08a: Herbert Sauro just finished his opening remarks

9:10a: Raik is explaining modularity and hierarchies in life:

  • atoms > residues > domains > proteins > assemblies > organelles > cells >
  • Whenever we learn to control and manipulate one of these layers, we learn a lot about the next layer.
  • Natural systems have evolved into functional modules, but ones that often integrate with each other in practicality (protein domains interacting with one another)
9:12a: SB Dogma:
  • Standardization
  • Decoupling
  • Abstraction
  • Free Exchange (of modules)
9:20a: Overview of BioBricks: A Standard for assembling DNAs in an idempotent way developed by Tom Knight.
9:21a: What’s up with BBa (aka BB1.0)?  BBa is not good for doing protein fusions - the scar introduces a frameshift.  BioFusion (Silver Lab) standard fixes the frameshift, but it has some issues anyway (No start or stop codons).  Freiburg Format is better, includes start and stop codons.  Also see BioBricks++ (Austin Che?; MIT) and BioBrick Extreme (Jonathan Golar, JC Anderson; Berkeley).
9:25a: Data Exchange: What’s a BioBrick, in a more abstract sense?  Each one has a unique DNA sequence. They can be composed of multiple other BioBricks, as long as the unique DNA sequence requirement is satisfied.  What about a Device?  One or more BioBricks working together that implement a functional interface.
9:29a: Minimal BioBrick DataModel! ID, DNA, Format (aka construction standard) description… see more at the BBF:TechnicalStandards
9:30a: BioBricks should be categorized into non exclusive families (multiple inheritance). [How is the category ontology organized?]
9:38a:Don’t bother with web services - build a web of registries out of SB catalogs that support a RESTful way of asking for data.
11:32a: John Cumbers on DeLaRAP: Dual Luciferace Assay for RBSs And Promoters.  Main advantage over Jason’s measurement standard: it’s much more sensitive, because it produces photons, instead of fluorescence.  Main use: characterizing parts on low-copy plasmids.
11:45a: Defining a standard promoter.  Why?
  • want gene expression to be correlated to promoter strength
  • ensure all biobrick promoters at the same 5′UTR to mRNA
  • remove secondary effects such as mRNA stability
  • in general, ensure same expression of different CDS with the same promoter
Promoter Standard should enforce these three criteria:
  1. Part spans from -100 to +20 nucleotides
  2. -35, -10 interacts with sigma70 subunit
  3. region between -10 and +1 is of the form: TATTATnnnnBCAT.

11:52a: We like to make analogies to electronic components.  They are nice models because the can be described in nice physical, readily measurable, absolute unit.  It’s not that they are always simple - a lot of the modules have nonlinear behaviors - but we can make really good predictions about them.

Currently, we are able to produce the right general architecture of devices: three promoters that inhibit eachother serially -> a ring oscillator.  But actually figuring out the sequence that implements this successfully is incredibly hard.

RELATIVE STANDARDS + STANDARD COMPOSITE.  The standard composite is PROMOTER, RBS, CDS, TERMINATOR. Each part gets a behavior measurement in units relevant to that type of the part, as well as some kind of Standard Composite score that embodies how accurate the part measurement is in different compositions.

Another idea: STANDARD ANALYSIS.  Standard set of algorithms specific to particular device types that help predict the sequences of a small number of functionally modulated parts

Another idea: STANDARD SET OF COMPOSITIONS.  Analogy to standardized search for protein crystallization conditions. Standardizing Protein Crystallization - it’s black magic, right?  Wrong!  It’s been automated for about 7 years now!  You set up 50 crystallization conditions, screen for success, and the do it over again based on the initial results.  This is a dynamic search for the best crystallization conditions.

We need design standards to minimize the time, cost, and uncertainty with identifying a sequence conferring a specific quantitative phenotype.  They have got to be validated (i.e. it will be useful in 80% of the conditions).  And data models for documenting quantitative standards must reflect the form of validated design methods.

About this entry

You’re currently reading “SSSB Liveblog: Saturday Morning,” an entry on Bricklet

Published:
04.26.08 / 12pm
Category:
blog