|
NCI Anti-Cancer Drug Data Analysis
|
Analyses and Visualizations Using Neural Nets
|
|
The group of John Weinstein at NCI has carried out an analysis of
the GI50 (concentration at which 50 % growth inhibition is observed) values
of 141 chemical agents of known Mechanism Of Action (MOA) screened
individually on 60 cells lines. The GI50 values were used to classify the
141 chemicals into their appropriate MOA (out of 6 possible) using a neural
net classifier. The neural net classification accuracy achieved was 91.5%.
In the analysis and visualizations presented below, we show a similar neural
net classification accuracy using the GI50 data on Clementine. We also
demonstrate visually the clustering of the six MOA groups using the GI50
and chemical property descriptor databases.
|
|
|
Figure 1.
NCI Anti-Cancer Cell Data - 60 Cell Lines Predicting MOA
This image shows a visualization using Circular Parallel Coordinates
of the NCI Anti Cancer Data (131 points) from the
NCI
Anti-Cancer Agent Mechanism Database. The GI50 data for 131 chemical
compounds for each of 60 cancer cell lines is compared for 6 Mechanisms of
Action (6 different colors). This visualization has been called various
names such as "Star Plots" or "Sun Ray Plots". Often a single "Star Plot" is
shown as an icon, but in this case we superimpose 131 "Star Plots" on
one diagram.
|
|
|
|
LINK 1:
Visualization grand tour (15 Mb mpg file of 1402 frames)
|
|
LINK 2:
NCI anti-cancer drug data visualization.
|
|
|