RWE072605

From OpenWetWare

Jump to: navigation, search

Replication w/ Error Study Group

Notes from July 26, 2005 meeting

  1. DE introduced problem from engineering perspective & provided naive mapping into information theory (i.e., parents = sender, offspring = receiver, machine design = message, replication process = channel, and so on).
  2. TK provided a tutorial on parity
    1. idea of adding a parity bit to a message
    2. idea of Hamming codes & logarythmic encoding
    3. idea of feedforward correction (information sent to enable both message checking and correcting)
    4. idea of feedback correction (only information needed for error detection sent with message, but channel in place for receiver to request a resend if needed)
      1. discussion of tradeoffs between feedforward and feedback correction
  3. more discussion of possible framework (see figure below)
  4. discussion of folks interests and personal goals
    1. DE -- design engineered biological systems according to this specification...
    2. JG -- how atrophy busting mechanisms work & even arise via evolution in the first place???
    3. JK -- how does the balance between persistence and adaptation work?
    4. GJS -- same as DE
    5. BC -- wants statistic that defines performance stability (also something about coupling of system function and objectives)
    6. TK -- wants ability to kill cells when we need them killed (i.e., how to responsibly engineer and release critters?)
    7. AL -- wants to be able to engineer systems that continue to function in a changing environment!
    8. AC -- how to implement information processing codes? E.g., hope to keep a non-function piece of DNA around (w/o taking the easy way out!)
  5. discussion of how codon space could be reclaimed to make it really fail fast (i.e., only use 20 coding codons, with the remaining 44 coding stops). points made that 64 codons don't provide enough space in codon space, and that math means that you need to go to 4-base codons for full coverage of fail fast (where fail fast is defined as all single point mutations lead to a stop).
  6. DE mentions Haussler paper, on list for next meeting.
  7. JK also wanting (willing) to talk about his thesis proposal (given that he passed, et cetera).

Image:RWEFramework1.jpg PPT file for above image

Personal tools