R Notebook

library(data.table)
library(tidyverse)

genes_merged <- read.csv("../data/merged_genes_full.csv",row.names = 1, header= TRUE)
genes_merged <- column_to_rownames(genes_merged,var="gene_names")

ref_gene <- as.numeric(genes_merged["acnA",])

# gene_count =as.data.frame(rowSums(genes_merged))
# gene_count <- tibble::rownames_to_column(gene_count, "gene_name")


# Calculate Jaccard Similarity
gene_matches <- colSums(t(genes_merged)+ref_gene==2)
gene_totalhits <- colSums(t(genes_merged)+ref_gene>0)
fave_genes <- sort(gene_matches/gene_totalhits, decreasing = TRUE)

# Check the confusion matrix of the top 10 best matched genes
test <- lapply(names(head(fave_genes, 10)), function(gene_name){
  table(as.numeric(genes_merged["acnA",]), 
        as.numeric(genes_merged[gene_name,]))
})

# plot the ordered data based on similarity score
x_steps = seq(1, length(fave_genes)-1, 1)
y_steps = fave_genes[-1]
df = data.frame(x=x_steps,y=y_steps)
ggplot(data=df,mapping = aes(x, y))+geom_point()

xlab("Rank Order")

$x
[1] "Rank Order"

attr(,"class")
[1] "labels"

ylab("Similarity Score")

$y
[1] "Similarity Score"

attr(,"class")
[1] "labels"

nifH is kinda interesting - lots of metal binding genes associated amidst the other nif genes