10:00 am – 5:00 pm, 7:00 pm – 11:40 pm
Literature
- interesting article on FACs smFISH sequencing from vanO group.
- interesting article on noise models for single cell sequencing from vanO group as well
STORM
- D01 finished smoothly
- setting up to re-image E11-F01 (not sure why these movies look weird or what happened to the conventional images)
- new E11-F01 images look much brighter (?). Not all movies finished analyzing.
Simulations
- simulating with
- mild compression force,
- 500 nodes
- periodic 25 node attractive force
- rigid body exclusion force, no angular forces
* repulsive force seems to be too strong / integrator steps too large — polymer exhibits weird fly apart behavior at certain points in time.
2. removed compression force
– overall volume becomes much greater in more frames
– system still exhibits extreme changes in volume.
3. removed attractive force. Only Harmonic force, rigid body repulsive force, remain.
4. Problem seems to have been missing parentheses in attractive force (and possibly some sensitivity to the number of nodes, prefers 51 to 50 or 201 to 200. Not rigerously verified this part though). Wrote matlab script to update number of nodes and simulation time and call python scripts.
5. Now running on sizes 51:25:1000, with same periodic 25 node attractive force, no compression, rigid body exclusion, no angular forces. Will hopefully finish computation sometime tonight or tomorrow morning.
– nodes are still explosive — extreme changes in volume with attractive force
– changing timestep and reducing sharpness (powers) on repulsive and attractive forces may have helped a little but not much
– offsetting the minimum of the attractive force and the molecule edge seems to have helped a little bit. (probably should do this by a full molecule diameter).
Additional technical observations
- can only switch user for a short period of time before the gpu stops computing
- should try to get this to run on Monet.
Discussion
- are structural domains independent of and pre-existing to chromatin protien domains (e.g. H3K9me3 domains reported by Dixon et al).