Reverse edge direction in `as_tidy_dagitty()` output

[gomesleduardo]

Hi, guys!

Description

I encountered an issue with the as_tidy_dagitty() function while following the example in the documentation. The resulting
DAG object appears to store the relationships correctly, but the data tibble has the edges defined in the opposite direction.

Reproducible Example

library(ggdag)
dag <- data.frame(name = c("c", "c", "x"), to = c("x", "y", "y")) %>% 
  as_tidy_dagitty()

print(dag$data)
print(dag$dag)

Observed Behavior

Output of dag$data:

# A tibble: 4 × 8
  name       x       y direction to      xend    yend circular
  <chr>  <dbl>   <dbl> <fct>     <chr>  <dbl>   <dbl> <lgl>   
1 c     -0.572  0.911  ->        x      0.420  0.765  FALSE   
2 c     -0.572  0.911  ->        y     -0.203 -0.0206 FALSE   
3 x      0.420  0.765  ->        y     -0.203 -0.0206 FALSE   
4 y     -0.203 -0.0206 NA        NA    NA     NA      FALSE   

Output of dag$dag:

dag {
c [pos="-0.572,0.911"]
x [pos="0.420,0.765"]
y [pos="-0.203,-0.021"]
x -> c
y -> c
y -> x
}

# ARCS ARE REVERSED!

Environment

# R Version: 4.3.1
# ggdag Version: 0.2.13
# dagitty Version: 0.3-4

Would appreciate any guidance or fixes for this issue.

Thanks for your work on this package!

[malcolmbarrett]

This may already be fixed in the development version if memory serves correctly. Could you try installing the version on GitHub (see README if you need instructions) and reporting back?

[malcolmbarrett]

Confirmed this is still the case in dev

[malcolmbarrett]

This is specifically a bug in as_tidy_dagitty.data.frame(). Other methods work correctly:

library(ggdag, warn.conflicts = FALSE)

dag <- dagify(x ~ c, y ~ x + c) |>
  tidy_dagitty()

dag$data
#> # A tibble: 4 × 8
#>   name       x      y direction to      xend   yend circular
#>   <chr>  <dbl>  <dbl> <fct>     <chr>  <dbl>  <dbl> <lgl>   
#> 1 c     -0.248  0.594 ->        x     -0.717 -0.283 FALSE   
#> 2 c     -0.248  0.594 ->        y      0.277 -0.251 FALSE   
#> 3 x     -0.717 -0.283 ->        y      0.277 -0.251 FALSE   
#> 4 y      0.277 -0.251 <NA>      <NA>  NA     NA     FALSE
dag$dag
#> dag {
#> c [pos="-0.248,0.594"]
#> x [pos="-0.717,-0.283"]
#> y [pos="0.277,-0.251"]
#> c -> x
#> c -> y
#> x -> y
#> }

^{Created on 2025-02-03 with reprex v2.1.1}

[malcolmbarrett]

I said earlier this was still the case in dev but I actually can't replicate it now. Could you try again?

> library(ggdag)
dag <- data.frame(name = c("c", "c", "x"), to = c("x", "y", "y")) %>% 
   as_tidy_dagitty()
 
print(dag$data)
print(dag$dag)
# A tibble: 4 × 8
  name        x      y direction to      xend   yend circular
  <chr>   <dbl>  <dbl> <fct>     <chr>  <dbl>  <dbl> <lgl>   
1 c     -0.0568 -0.604 ->        x     -0.730  0.139 FALSE   
2 c     -0.0568 -0.604 ->        y      0.250  0.350 FALSE   
3 x     -0.730   0.139 ->        y      0.250  0.350 FALSE   
4 y      0.250   0.350 NA        NA    NA     NA     FALSE   

dag {
c [pos="-0.057,-0.604"]
x [pos="-0.730,0.139"]
y [pos="0.250,0.350"]
c -> x
c -> y
x -> y
}