Changes between Version 3 and Version 4 of 2024-6-6


Ignore:
Timestamp:
Jun 6, 2024, 11:52:50 AM (17 months ago)
Author:
Zach Pearson
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • 2024-6-6

    v3 v4  
    1010
    1111= Discussion Notes =
    12 
     12* Wynton upgrades
     13  * UCSD got money for GPUs but other UCs can't use them
     14    * they're hosted at SDSC which has painful security requirements
     15  * Wynton is getting two new GPU nodes that the RBVI will get access to
     16    * 8 x L40 48GB GPUs to be added to preexisting A40s
     17    * Officially NRNB nodes but RBVI is a member of NRNB
     18    * UCSD and Toronto people may also use them
     19  * Plato is also getting a memory upgrade (2 nodes go up to 1TB)
     20* 1.8 release
     21  * Nearly ready, Eric needs to track down tricky GC bug
     22* Tom gave LookSee demo with his wife, was successful
     23  * New intern, in the middle of finals, may be here next Thursday
     24* Elaine presentation
     25  * Embedding Residue Interaction Networks
     26    * No training required, uses autoencoder
     27    * Handles many time points efficiently
     28  * Motivation
     29    * Hard to get meaning from MD trajectories, many variables
     30    * Typically need expert knowledge to know which variables to pay attention to
     31    * RINs capture structural changes on many scales
     32    * Prior analyses used averages or only a few snapshots, lost temporal information
     33  * Implementation
     34    * Edge drawn b/t non-sequence-adjacent residues with any atoms within 6 angstroms
     35    * residue closeness centrality --> N-dimensional feature vector for each time point
     36  * Small Protein Example: Trp-Cage (20 residues)
     37    * Super long simulation (208 microseconds, >1M frames)
     38  * Supp data: Trp-Cage PCA and UMAP
     39    * Author's note: compared to EncoderMap, PCA more global, UMAP more local
     40  * Multidomain Example: FAT10 (165 residues)
     41    * FAT10 has two ubiquitin-like domains wiht a flexible linker and tails
     42    * 50 simulations, 50ns
     43    * Authors hypothesize that different closed conformations offer different interaction sites for FAT10's many binding partners
     44    * Map resolves functionally different conformations of a complex system over time, yet retains an interpretable global organization
     45  * Alternative Featurizations
     46    * Adjacency matrix colored by density or R_g
     47    * Degree centrality colored by density or R_g
     48    * Backbone dihedral angles colored by density or R_g
     49  * Discussion
     50    * Protein graph can be defined at different resolutions/with different criteria
     51    * RINs cannot distinguish fine local changes, best for systems that undergo strong changes
    1352= Action Items =