Saturday 10/18/14

10:00 am – 1:00 pm, 7:00pm – 9:00pm,

Fly work

  • flip stocks

Simulations

  • troubleshooting running polymer simulations on Odyssey GPU
  • issues using polymer load and save commands — requires joblib
  • testing out with Anaconda via sbatch command
Posted in Summaries | Comments Off

Friday 10/17/14

10:00 am – 11:00 pm

Project 2

  • see post

Chromatin project

  • working on manuscript
  • playing with chromatin simulations
Posted in Summaries | Comments Off

Thursday 10/16/14

10:00 am – 12:45 am

Chromatin Modeling

initalization

  • compressed random walk polymer maybe not best initial fractal globule state
  • try using a sticky polymer, these have pretty good fractal structure all the way down.
  • too sticky polymer settles down too slowly (even with local energy minimization).
  • starting from this condensed state and letting the system relax rapidly back to the new target density is also bad
    • in particular, the black chromatin rebounds faster than the yellow, and the yellow stays compressed.

Initial condition variation

  • starting on the same section of the same initial fractal globule has strong effects on behavior
  • fractal globules having all scales of loops have dramatic variation in the size of any given domain. It seems to me insufficient attention has been given to this variability in the literature.

effects of Energy minimization

  • energy minimization greatly helps compacting blue domains
  • energy minimization possibly expands the black domains at the expense of the softer yellow domains.
  • maybe if the yellows are sufficiently transparent they won’t get compressed?
  • not sure if this true — maybe just an artifact of the fact my circ permute domains got all my domains mixed up.

Debugging

  • okay, debugged new permute domains. hopefully this works in disrupting the dependence on initial condition.
  • things looking a bit more promising. Need more averaging!

WholeNucSimNoAve

WholeNucSim2

Posted in Summaries | Comments Off

Wednesday 10/15/14

Chromatin

Model revisions

  • Add sticky and non-sticky blue (so not all nodes are sticky)
  • helps blue mixing

updates to chromatin model

largeBlueAndBlack smallBlueAndBlack SomeBlueDomains NucleusImage YandKimage6 YandKimage5 YandKimage4 YintScaling SimulatedYintactScaling YandKimage3 YandKimage2 YandKimage blueInternalScalingSimulation2 blueInternalScalingSimulation blueIntactScalingSimulation

working on Polymer simulations

WholeNucSim

PolymerForV5sim

withPartStickyBlue

FractalGlobuleScalingN15000

Posted in Summaries | Comments Off

Monday 10/13/14

10:00 am – 2:45 am

Chromatin

Literature

  • reading Mirny 2011 review

Simulations

Data for Bogdan

  • send volume table to Bodgan to explore densities
  • send table of internal volumes
  • send table of intact Rgs
  • send table of internal Rgs

To plot for XZ meeting

  • blue internal scaling, just BX-C data. With .5 and .33 reference lines
  • blue internal scaling, just ANT-C data. With .5 and .33 reference lines
  • simulation of pure black chromatin.

Simulations

  • got black working
  • added yellow – too many holes? allows black to equilibrate?
  • reduced yellow – seems to help, hard to tell when not a lot of yellow to go by.
  • went back to all black – back to 0.4 scaling?
  • went back to all black with Grosberg repulsion, still 0.4 scaling.
  • reduced number of domains and increased min domain size (maybe the too smalls are getting in the way).
  • this fixes things, back to 0.3. ‘YellowBlackAllK4_Rnd’
  • switching back from Grosberg to tagged force. Seems to be okay. 0.32 ‘YellowBlackAllK5_Rnd’
  • adding back 10% yellow ‘YellowBlackAllK6_Rnd’. Black seems to stay down at 0.3 to 0.35, though the data is supernoisy with this few points. Yellow with 3 points is useless to fit. showing 0.24 (-1,1.5)
  • ‘YellowBlackAllK7_Rnd’ with more yellow poitns is still pretty even (0.32 black [.21 .42], 0.26 yellow [.13 .4])
  • let’s try making yellow a bit more phantom than 5:50. ‘YellowBlackAllK8_Rnd’ looking at 2::50 comparison
  • This looks a bit more promising, black is staying at .3, yellow is occassionally climbing above (though still very noise).
  • running 200 step, more subchains, larger polymer, long run overnight ‘YellowBlackAllK11_Rnd’.
  • overnight run looks decent. data is still very noisy, maybe this will average away if we use a random chain configuration to start instead of the deterministic wrapping, and we do we enough repeats. (current system of repeats doesn’t average away noise).

Project 2

  • restart Cajal running on scans
Posted in Summaries | Comments Off

Sunday 10/12/14

11:00 am – 11:00 pm

Computation

  • investigate local data running on Odyssey CPU
  • does not look straight-forward. Writing to RChelp for more info

Project 2

  • run threshold scans on Cajal — now going on original test dataset and new dataset

Chromatin

  • new simulation, stiffer blue, shorter run time, less chain crossing
  • black chromatin still gets around
Posted in Summaries | Comments Off

Saturday 09/11/14

10:00 am – 8:00 pm

Modeling Calculations

fractions of chromatin types in genome

  • of the 120 Mb of classified sequence (some 60 Mb ‘other’ class)
  • estimate: from Pc track, 1 – 3% ‘Blue’
  • estimate: from yellow/red Filion data: 34% ‘Yellow’ (21% ‘Yellow’,13% ‘Red’).
  • estimate: from black and blue minus true blue: 64% ‘Black’

estimate monomer length and size

I estimate the intersection of three scaling laws to occur between 0.4 kb and 3.2 kb to ~95% confidence*.

*About the calculation:
I have 3 lines from our 3 scaling laws. Around each line I can plot the lower and upper uncertainty bounds and turn each line into a pair of lower and upper extema. If I compute the first time the lower extrema of the yellow crosses the upper extrema of the blue I call that the largest value possible for the intersection. But I think its wrong to compute this using the 95% bounds, because the probability all 3 of these datasets are at or outside their 95% limit is (0.05)^3 not (0.05). So instead I use the 63% bounds around the line and compute the earliest (and latest) time where they could all intersect and I get the above numbers.

The volume of the chromatin at the crossing point depends a little which crossing point in this range we look at.

  • at 0.4 kb it ranges from 10^5.4 to 10^5.8 nm
  • at 1 kb it ranges from 10^5.8 to 10^6.1 nm
  • at 3.2 kb it ranges from 10^6.3 10^6.7 nm
    based on the the confidence intervals.

Any one of the monomer lengths (0.4 to 3.2) can satisfy the constraint that the biggest blue has a volume fraction less than 1.0 if we chose its lower bound for the the monomer volume estimate. (the upper estimates all put the volume fraction way above 1).

Modeling to do

  • check convergence time:
    • run mulitple very long simulations, see how Rg changes at several subchain lengths for blue, black and yellow.
  • formalize computation of monomer size. Done. See above.
Posted in Summaries | Comments Off

Protected: chromatin modeling results

This content is password protected. To view it please enter your password below:

Posted in Chromatin | Comments Off

Friday 10/10/14

9:45 am – 12:15 am

Project 2

  • see post

Chromatin

Openmm configuration

  • updated openmm-polymer in Python26/Lib/site-packages/ to latest download
  • archived old openmm-polymer in Research/Software/OpenMMchromatin/Archive (this had some of my scripts as well)
  • reorganizing OpenMMchromatin into
    • Archive
    • Templates
    • Simulations
    • Functions
    • Scratch

Coding with Bogdan

  • Realization: should only look at scaling much shorter than constrained volume traversal.
    • the flat scaling in fractal globules is just after the polymer reaches the boundaries of confinement
    • sticky polymers set their own boundaries based on the size of the ball in which the nodes of the polymer still have sufficient contact.
  • implemented variable stiffness simulation
  • implemented variable stickiness simulation

Chromatin simulation results

Posted in Summaries | Comments Off

Protected: project 2 update: 10/09/14

This content is password protected. To view it please enter your password below:

Posted in Genomics | Comments Off