Table 4. Constructing networks of phenotypic neighbors
|
Construct MCs as feature vectors: MC → 8-dimensional integer vector
|
| 4.1 |
Load auxiliary module featureMap. |
| 4.2 |
Load main module featureVectors.
Arguments:
Tabular data, column to use as key for each vector and names of features.
Algorithm outline:
Call featureMap to get a mapping which assigns each PAM state found to an integer
value. For each element in the key column the set of features (third argument) is substituted
according to the map previously obtained.
Returns:
Feature vectors and mapping used. |
| 4.3 |
Assign to featureList a list of PAMs to use as they appear in the header of AF3A and run
{fvs, fm} = featureVectors[excelAF3A, "Species", featureList].
|
| 4.4 |
Remove duplicate MCs from fvs to get a unique list of 76 feature vectors.
|
|
Join MCs in a network according to PAM state changes
|
| 4.5 |
Load auxiliary module similarityMatrix. |
| 4.6 |
Load main module buildDistanceGraph.
Arguments:
List of unique feature vectors of the same dimension. Boolean value to specify whether to stop after
a first pass or continue until joining all vertices
Algorithm outline:
Use similarityMatrix in order to get a square matrix with the number of differences
of all pairwise comparisons of the vectors. Starting with each vector as a one-vertex connected
component, compare it with the vertices of the other components, joining two nodes if the number of
differences is 1. Proceed equally with the rest of the connected components. Increase the number of
differences until all nodes are in one component.
Returns:
List of edges. |
| 4.7 |
Load auxiliary module addVirtualNodes.
Algorithm outline:
For each edge in the feature vector distance graph, insert nodes proportionally to the number of
differences between all nodes that differ in more than one change.
|
| 4.8 |
Plot graph returned by addVirtualNodes[buildDistanceGraph[Sort[Union[fvs]],True]]
with visualization modules setEdgeColor and setVertex.
|
| 4.9 |
Plot graph returned by addVirtualNodes[buildDistanceGraph[Sort[Union[fvs]],False]]
with visualization modules setEdgeColor and setVertex.
|
| 4.10 |
Load auxiliary module addMultiLabeledEdges.
Algorithm outline:
For each edge joining two feature vectors (2 or 3 in the case of more than one change between them)
add as many edges as different lineage changes share the MC. Label the edges accordingly.
|
| 4.11 |
Plot graph returned by
addMultiLabeldEdges[addVirtualNodes[buildDistanceGraph [Sort[Union[fvs]],False]]]
with visualization modules setEdgeColor2 and setVertex.
|
Table 5. Phenotypic neighbors network analyses
|
Computation of degree distribution and average clustering
coefficient of phenotypic neighbors networks
|
| 5.1 |
Obtain adjacency matrix |
| 5.2 |
Transform graph into Combinatorica format from adjacency matrix |
| 5.3 |
Load module degreeDistribution.
Arguments:
Graph (in Combinatorica format)
Returns:
Probability degree distribution,
P(k)
, and cumulative probability degree distribution,
Pc(k)
for all degrees
k
in input graph.
 - Pc(k).png)
Where N(i) → Nodes of degree
i ;
N → Number of nodes ;
K → Maximum degree |
| 5.4 |
Run degreeDistribution(graph). |
| 5.5 |
Load auxiliary module clusteringCoefficient.
Arguments:
Graph (in Combinatorica format) and node number
Returns:
Clustering coefficient of specified node,
C(n)
.
.png)
Where
δ(ni) →
Degree of node ni ;
μ(ni) →
Number of edges between distance 1 neighbors of ni
|
| 5.6 |
Load main module averageClusteringCoefficients.
Arguments:
Graph (in Combinatorica format).
Returns:
Clustering coefficient
C(k)
for all degrees
k
in input graph.
.png)
|
| 5.7 |
Run averageClusteringCoefficients(graph). |
| 5.8 |
Calculate the mean shortest path length .png) :
Run Mean[Mean[AllPairsShortestPath[graph]]]
|