How to create a vector from elements in common in vectors R

Question

I have several character vectors of genes containing names of the species in which they're found, and I made an UpSetR plot to show the number of species in common across genes. Now I'd like to do the opposite: Plotting the number of genes in common across species, yet I don't know how to do it.

Example of what I have:

gene1 <- c("Panda", "Dog", "Chicken")
gene2 <- c("Human", "Panda", "Dog")
gene3 <- c("Human", "Panda", "Chicken")  
...#About 20+ genes with 100+ species each

Example of what I would like to have as a result:

Panda <- c("gene1", "gene2", "gene3")
Dog <- c("gene1", "gene2")
Human <- c("gene2", "gene3")
Chicken <- c("gene1", "gene3")
...

I know it is conceptually easy, yet logistically more complicated. Can anyone give me a clue?

Thank you!

score 8 · Accepted Answer · answered Mar 19 '18 at 19:27

You can use unstack from base R:

unstack(stack(mget(ls(pattern="gene"))),ind~values)
$Chicken
[1] "gene1" "gene3"

$Dog
[1] "gene1" "gene2"

$Human
[1] "gene2" "gene3"

$Panda
[1] "gene1" "gene2" "gene3"

You can end up listing this to the environment by list2env function

Breakdown:

 l = mget(ls(pattern="gene"))#get all the genes in a list
 m = unstack(stack(l),ind~values)# Stack them, then unstack with the required formula
 m
$Chicken
[1] "gene1" "gene3"

$Dog
[1] "gene1" "gene2"

$Human
[1] "gene2" "gene3"

$Panda
[1] "gene1" "gene2" "gene3"

 list2env(m,.GlobalEnv)
 Dog
 [1] "gene1" "gene2"

score 3 · Answer 2 · answered Mar 19 '18 at 19:25

First of all I think for most purposes it's better to store gene vectors in a list, as in

genes <- list(gene1 = gene1, gene2 = gene2, gene3 = gene3)

Then one base R approach would be

genes.v <- unlist(genes)
names(genes.v) <- rep(names(genes), times = lengths(genes))
species <- lapply(unique(genes.v), function(g) names(genes.v)[g == genes.v])
names(species) <- unique(genes.v)
species
# $Panda
# [1] "gene1" "gene2" "gene3"
#
# $Dog
# [1] "gene1" "gene2"
#
# $Chicken
# [1] "gene1" "gene3"
#
# $Human
# [1] "gene2" "gene3"

genes.v is a named vector of all the species with the genes being their names. However, when to species have the same, e.g., gene1, then those names are gene11 and gene12. That's what I fix in the second line. Then in the third line I go over all the species and create the resulting list, except that in the fourth line I add species names.

score 3 · Answer 3 · answered Mar 19 '18 at 19:49

Put the data in a list, to begin with. That makes it easier to work with.

genes <- list(
    gene1 = c("Panda", "Dog", "Chicken"),
    gene2 = c("Human", "Panda", "Dog"),
    gene3 = c("Human", "Panda", "Chicken")
)

Then we can get the species names from there.

species <- unique(unlist(genes))

With this data

> species
[1] "Panda"   "Dog"     "Chicken" "Human"

For each of these, we want to check if the name is contained in a gene. That is a job for Map (or its cousin lapply, but I like Map):

get_genes_for_species <- function(s) {
    contained <- unlist(Map(function(gene) s %in% gene, genes))
    names(genes)[contained]
}
genes_per_species <- Map(get_genes_for_species, species)

Now you have a list of lists, one list per species, containing the genes found in that species.

> genes_per_species
$Panda
[1] "gene1" "gene2" "gene3"

$Dog
[1] "gene1" "gene2"

$Chicken
[1] "gene1" "gene3"

$Human
[1] "gene2" "gene3"

score 1 · Answer 4 · answered Mar 19 '18 at 19:23

1

You can try this.

gene  <-unique(c(gene1,gene2,gene3))
TF    <-data.frame(Species = gene)

TF$gene1 <- gene%in%gene1
TF$gene2 <- gene%in%gene2
TF$gene3 <- gene%in%gene3

> TF
  Species gene1 gene2 gene3
1   Panda  TRUE  TRUE  TRUE
2     Dog  TRUE  TRUE FALSE
3 Chicken  TRUE FALSE  TRUE
4   Human FALSE  TRUE  TRUE

answered Mar 19 '18 at 19:23

Mohammad Tanvir Ahamed

85
4

That seems useful too! I was thinking now in creating a 1-and-0 matrix (species in rows, genes in cols, just like yours) from a list containing my vectors, do you know how to do it? Thanks! – Guillermo Reales Mar 19 '18 at 19:57
1

@GuillermoReales `table(reshape2::melt(l))`, where `l` is your list. – Henrik Mar 19 '18 at 20:10
@Henrik I managed to do it eventually using `mtabulate` function from `qdapTools` package. In case it'd be useful from someone, it does the trick too. – Guillermo Reales Mar 19 '18 at 20:20
And from `d <- reshape2::melt(l)` you could easily create the 'transposed' list: `split(d$L1, d$value)`. – Henrik Mar 19 '18 at 20:26

Nettle · Answer 5 · 2018-09-16T19:03:32.663

Here's a variation that embraces the tidyverse and puts the result in a neat dataframe.

The trick is to concatenate results with str_c and summarise.

   tibble(gene1 = gene1, 
          gene2 = gene2, 
          gene3 = gene3) %>% 
   gather(gene_name, gene_type) %>% 
   group_by(gene_type) %>% 
   summarise(genes = str_c(gene_name, collapse = ", "))

# A tibble: 4 x 2
  gene_type genes              
  <chr>     <chr>              
1 Chicken   gene1, gene3       
2 Dog       gene1, gene2       
3 Human     gene2, gene3       
4 Panda     gene1, gene2, gene3

I agree with Julius (above) that best way to store gene vectors is with a list. A named list would be even better, as:

my_gene_list <- set_names(list(gene1, gene2, gene3), str_c("gene", 1:3) )

This would neatly produce the same result...

 my_gene_list %>% as_tibble() %>% 
   gather(gene_name, gene_type) %>% 
   group_by(gene_type) %>% 
   summarise(genes = str_c(gene_name, collapse = ", "))

# A tibble: 4 x 2
  gene_type genes              
  <chr>     <chr>              
1 Chicken   gene1, gene3       
2 Dog       gene1, gene2       
3 Human     gene2, gene3       
4 Panda     gene1, gene2, gene3

How to create a vector from elements in common in vectors R

5 Answers5