Error in seq_len about amr

[msberends]

Hi! Yes, sure. It seems to be a bug, I’ll look into it!

[susheelbhanu]

Thank you

[msberends]

It was indeed a bug, I'll push a fix to this repo today.

But even fixed, the text used as input might not be very suitable for retrieving antibiotic codes - in what you posted there are almost no antibiotics to be found if I manually scan through them with my own knowledge about antibiotic/antifungal drugs.

This also leads to many false-positives. For example, the problematic one that led to the error was nr 17:

[17] "Euthyrox 150µg [150µg Levothyroxin natrium, (145,88µg Levothyroxin)] 1x1 => Hypothyreose seit 1980"

Without the bug, this would be the result:

ab_from_text(x[17], type = "drug", thorough_search = TRUE)
#> [[1]]
#> Class <ab>
#> [1] INH AKF AKF AKF CPO

So that's a list of isoniazid, 3 times amikacin/fosfomycin, and cefpirome. Not very helpful I guess... The ab_from_text() function can only retrieve antibiotic/antifungal drugs, not other drugs.

Does this clarify things a bit?

[susheelbhanu]

Dear @msberends,

Thank you very much for pushing this fix.

And indeed what you say makes sense. I have a list of 300 unique drugs, and was using the package to reduce my workload of manually validating which ones are antibiotics and which are not 😉 . It was the first step of my screening process, or else I'd have to look through entire list manually and confirm via a Google search.

Any (false)-positives that we get from this analyses, we will manually verify to make sure they are indeed antibiotics/antifungals, so this is already helpful.

I am assuming though, that if it doesn't find any matches, it'll simply report a NA - is that correct?

And thanks again for fixing it so quickly.

[msberends]

That’s a fine workflow!

Indeed, when no results are found it will return NA.

[susheelbhanu]

Awesome, thank you. It is a nice workflow indeed 🙂.