

Line 1: 
Line 1: 
  Back to [[Ty ThomsonTy Thomson's user page]]<br>
 
  Back to [[TMT Thesis ProjectTy Thomson's thesis page]]
 
   
  == Biology ==
 
 
 
  #Show that cells can live on a chip
 
  #*Stick yeast cells down in the channel, and flow media (at a slow rate) over them. Take a picture every 5 mintutes and compile into a movie of yeast cells growing (hopefully). Need to start with cells growing exponentially, and concentrate to OD 1.52. Try sonicating briefly to break up clumps (talk to Jeff).
 
  #*I've proven to myself that cells can grow on a chip, though I still don't have a good movie demonstrating this. I will try to get this evidence so everyone will be a believer.
 
  #Show that you can control in ON/OFF fashion response of cells to alpha factor
 
  #*Using strain with YFP driven by P<sub>prm1</sub> promoter. Show that cells won't react (ie fluoresce) when they are not in part of channel where alpha factor is flowing, and that they do react when they are exposed to alpha factor. I need to use casein in the media to block pheromone adsorption to the tubes and channel walls.
 
  #Find out if reset of receptor/G protein subsystem is limited by pheromone dissociation or Ste2 internalization.
 
  #*Hit cells with a short dose of pheromone and see if reset is on the order of 45 mintutes (internalization) or 10 minutes (dissociation). See if Alejandro has already done this.
 
  #(Is yeast pheromone response the best model system for this project?)
 
 
 
  == Data Collection/Analysis ==
 
 
 
  *Show that you can measure Ste5YFP translocation to membrane
 
  *This will involve either using or reimplementing the image analysis tools used by Alejandro and Andrew. Also, I might want to use/reimplement their autofocus routine. I should look into this soon.
 
 
 
  == Signal Design ==
 
 
 
  Figure out a way to measure the coupling between parameters for a given input timecourse
 
  *Look at the parameter sensitivities over time. I can use the correlation coefficient as a metic for the coupling between parameters. Define the Independence Coefficient as 1  abs(correlation coeff). This give a measure of the independence of each pair of parameters.
 
 
 
 
 
  Figure out how to select input timecourses that will allow for independent estimates of parameters
 
  *Here we need to pick from a number of possible experiments (input timecourses) to select a set of experiments that will allow for the best possible estimation of each parameter. Presumeably we are looking for experiments where two parameters are independent and the sensitivity is high for both parameters (so they can be simultaneously/independently estimated from one experiment), and experiments where the sensitivity is high for only one parameter out of a pair of nonindependent (ie correlated) parameters.
 
 
 
 
 
  *I don't know how parameter sensitivity is related to estimated parameter uncertainty. Is there necessarily any relationship?
 
  *Given a way to score experiments on their ability to simultaneously estimate two parameters, or estimate one parameter independently of another, I need to find a way to select a reasonable number of experiments that will maximize the information out of the parameter estimation.
 
  *Can parameter estimation be optimized in any way given knowledge of how parameters are correlated? If we're performing parameter estimation over mutliple experiments, can knowledge of sensitivity and correlation be used to optimize the process?
 
 
 
  == Parameter Estimation ==
 
 
 
  #Get jacobian working
 
  #Would knowledge of parameter groupings affect parameter estimation? I want to think about this some more, maybe chat with some people in the lab, and try to talk to John Tolsma (@Jacobian) about this.
 
 
 
  == Administrative Issues ==
 
 
 
  Thesis Committee<br>
 
  Q. Do we need a dynamic systems person on the committee?<br>
 
  *I don't know enough to be efficient at guiding myself through the parameter estimation and dynamic system analysis.
 