CCLE map: Multi-tiered genotype-to-phenotype network

Edges

Value: 70

Absolute value

Layers

Methylation

Copy number

Histone marks

miRNA

Expression

Protein

All

Genotype to phenotype
Starting layer:




nodes
borderWidth:
borderWidthSelected:
color
border:
background:
highlight
border:
background:
hover
border:
background:
opacity:
fixed
x:
y:
font
color:
size:
face:
background:
strokeWidth:
strokeColor:
hidden:
labelHighlightBold:
physics:
scaling
min:
max:
label:
shadow:
shape:
shapeProperties
borderDashes:
borderRadius:
interpolation:
useImageSize:
size:
edges
arrows
to:
middle:
from:
endPointOffset
from:
to:
arrowStrikethrough:
color
color:
highlight:
hover:
inherit:
opacity:
dashes:
font
color:
size:
face:
background:
strokeWidth:
strokeColor:
align:
hidden:
hoverWidth:
labelHighlightBold:
physics:
scaling
min:
max:
label
enabled:
min:
max:
maxVisible:
drawThreshold:
selectionWidth:
selfReferenceSize:
selfReference
size:
angle:
renderBehindTheNode:
shadow:
smooth
enabled:
type:
forceDirection:
roundness:
width:
layout
hierarchical:
interaction
dragNodes:
dragView:
hideEdgesOnDrag:
hideEdgesOnZoom:
hideNodesOnDrag:
hover:
keyboard
enabled:
speed
x:
y:
zoom:
bindToWindow:
autoFocus:
multiselect:
navigationButtons:
selectable:
selectConnectedEdges:
hoverConnectedEdges:
tooltipDelay:
zoomView:
zoomSpeed:
manipulation
enabled:
initiallyActive:
physics
enabled:
barnesHut
theta:
gravitationalConstant:
centralGravity:
springLength:
springConstant:
damping:
avoidOverlap:
maxVelocity:
minVelocity:
solver:
timestep:
wind
x:
y:
Building a genotype-to-phenotype map

CCLE map is a genotype-to-phenotype network that connects mutliple layers of omic networks. The network was built using data from the CCLE, DepMap, project ACHILLES,and PRISM projects that characterized several omic layers in more than 1700 cancer cell lines. Connecting the components of the network using Pearson Correlation Coefficient can result in spurious correlations that are not biologically meaningful. Approaches such Gaussian Graphical Models (GGMs) compute partial correlations to find direct regulatory interactions by conditioning the associations on all other variables of the system. In particular, computing partial correlations between two omic layers that have different structure can bias the estimation towards one or the other. Therefore, we built a multi-tiered genotype to phenotype map as a hypothesis generation tool using a novel method called DRAGON, that estimates partial correlations using a covariance shrinkage approach that allows to take into account disparate data structures in multi-omic data. CCLE map provides a data-driven resource that connects genotype to cellular pheontypes to mine biological associations in cancer cell lines.