在R中使用igraph为图添加属性时的节点匹配与度数异常问题

Let's work through your problems one by one to get your graph set up correctly:

1. Vertex Attribute Misalignment (Birthday Mismatch) #

The root cause here is the i + 1 in your loop when accessing the prop dataframe.

Why this happens #

• V(g) uses 1-based indexing, so the first vertex (i=1) corresponds to the first row of your adjacency matrix (node 27136).
• Your prop dataframe is ordered the same way: row 1 = node 27136, row 2 = node 27137, etc.
• By using prop[i + 1, j], you're pulling data from the next row in prop for each vertex, which shifts all attributes by one position.

The Fix #

Replace i + 1 with i, or even better, map attributes using vertex names to avoid issues if node order ever changes:

# First, make sure your graph vertices have the correct names from the adjacency matrix
V(g)$name <- rownames(m)

# Batch-set attributes using vertex name matching (safer than index-based)
for (attr_col in names(prop)) {
  V(g)[[attr_col]] <- prop[V(g)$name, attr_col]
}

2. Incorrect Node Degrees #

This is due to how graph.adjacency handles undirected graphs when your adjacency matrix isn't symmetric.

Why this happens #

When you use mode="undirected", igraph assumes your adjacency matrix is symmetric. If it's not, igraph only uses the upper triangle (i < j) of the matrix to create edges, ignoring all entries in the lower triangle. This is why your degree counts don't match the full matrix values.

The Fix #

You need to symmetricize your adjacency matrix first to ensure all undirected edges are accounted for. Here are two ways to do this:

Option 1: Manually Symmetricize the Matrix

# Add the matrix to its transpose to mirror edges
m_sym <- m + t(m)
# Remove self-loops (set diagonal to 0)
diag(m_sym) <- 0
# Create the undirected graph from the symmetric matrix
g <- graph.adjacency(m_sym, mode="undirected", weighted=NULL)

Option 2: Use mode="plus" (Automatic Symmetrization)

igraph's mode="plus" will automatically symmetricize the matrix by adding m and t(m) for you:

g <- graph.adjacency(m, mode="plus", weighted=NULL)

Full Corrected Code #

library(igraph)

# Load and process adjacency matrix
path <- "https://raw.githubusercontent.com/pranavn91/PhD/master/Expt/27022.csv"
dat <- read.csv(path, row.names=1, check.names=FALSE, header=T)
m <- as.matrix(dat)

# Create symmetric undirected graph
m_sym <- m + t(m)
diag(m_sym) <- 0
g <- graph.adjacency(m_sym, mode="undirected", weighted=NULL)

# Load and map vertex attributes
path2 <- "https://raw.githubusercontent.com/pranavn91/PhD/master/Expt/27022%20feat.csv"
prop <- read.csv(path2, row.names=1, check.names=FALSE, header=T)

# Set vertex names and map attributes
V(g)$name <- rownames(m)
for (attr_col in names(prop)) {
  V(g)[[attr_col]] <- prop[V(g)$name, attr_col]
}

# Verify fixes
# Check first vertex's birthday
cat("First vertex birthday:", V(g)[1]$birthday, "\n")
# Check degrees
cat("\nNode Degrees:\n")
print(igraph::degree(g, v=V(g)))

After these changes, your vertex attributes will align correctly, and the degrees will match your expected values: 27136-2, 27137-2, 27138-1, 27139-3.

内容的提问来源于stack exchange，提问作者pranav nerurkar