Challenges in Describing Signaling Networks

1
(No Transcript)
2
Challenges in Describing Signaling Networks
3
Cell Morphology Signaling State
4
Understanding how Signaling Networks that
Regulate Morphology are Organized and How
Information Flows through these Networks
RhoGTPases
5
Acquiring Morphological Signatures from Complex
Images
DAPIGFPF-Actin
6
Raw Morphological Data and Data Reduction
145 phenotypic features
7
Using Feature Graphs to Model Single-cell
Distributions
dsRNA y
4i,4j,-4k
Feature values i,j,k
2i,2j,-2k
8
Inference Based on Feature Graphs
This is the unknown signaling network we will
infer. For this slide, assume we know the
signaling network ahead of time
Intuition for Inference Based on Feature Graphs
Question What is the relationship between
feature graphs of genes in a signaling pathway?
F1, F2
RNAi Gene C
Expect feature graph to have a relatively small
number of edges Feature graph is approximately
the intersection of the feature graphs for RNAi A
and RNAi B
RNAi Gene A
F1, F2, F4
F1, F2, F3
Expect feature graph to have relatively large
number of edges
RNAi Gene B
Expect feature graph to have a relatively large
number of edges
9
Focus on Details of Feature Graph Construction
(future data sets will be larger)
For each FG, need to compute linear correlation
for C50 data points for all F(F-1)/2150149/2
pairs of features. Since there are N 250 TCs,
there are a total of NF(F-1)/2 linear
correlations to compute.
10
Focus on Details of Feature Graph Construction
For each FG, need to compute linear correlation
for C50 data points for all F(F-1)/2150149/2
pairs of features.
How to compute all pairwise correlations
efficiently?
matmul of FxC and CxF
Computation is dominated by matmul of FxC and CxF
matmul of Fx1 and 1xF
Matlab built-in corr does not work with ppeval
11
Parallelize in Dimension of TCs
Speed-up?
Parallel
Serial
12
What if the TCs Have Different Numbers of Cells
(C)?
13
What if the TCs Have Different Numbers of Cells
(C)?
Serial
Parallel
14
Summary and Conclusions

• Feature graph construction depends on
  computation of numerous linear correlations
• Parallelization was implemented
• But speed-ups were not realized (why not?)
• In fact, slower because of time required to move
  data to/from the server
• Speed-ups are realized for very large data
  sets because the server can handle larger data
  more smoothly than a typical PC. But this is not
  due to parallelization, rather due to hard drive
  usage.
• Why didnt parallelization result in gains in
  speed?
• Interactive Supercomputing doesnt preallocate
  matrices in Matlab
• Structure of problem?
• Coding?

15
Acknowledgments
Chris Bakal
Bonnie Berger
John Aach
Norbert Perrimon
George Church