odd errors with sequence length queries at ncbi_byname

[kdlafferty]

"ncbi_byname" works great most of the time (thanks!), but with limited testing fails about 5%. Which makes searching long lists of species cumbersome. It seems like there are some sequences in ncbii that lead to glitches, and these can be avoided by limiting the length ranges. But it is otherwise unclear why some of these queries fail consistently and other succeed or how to add options to get around it.

I should specify that my goal is to get a longish sequence by gene/species combination. across many species and many genes (for metabarcoding primer selection).

In Rstudio.
ncbi_byname(taxa="Coryphaena hippurus", gene = c("Coi"), seqrange = "1:2000")#fails
ncbi_byname(taxa="Coryphaena hippurus", gene = c("Coi"), seqrange = "500:750")#fails
ncbi_byname(taxa="Coryphaena hippurus", gene = c("Coi"), seqrange = "600:2000")#works
ncbi_byname(taxa="Sardinops melanostictus", gene = c("12s"), seqrange = "1:2000")#fails
ncbi_byname(taxa="Sardinops melanostictus", gene = c("12s"), seqrange = "1:1000")#works
ncbi_byname(taxa="Sardinops melanostictus", gene = c("12s"), seqrange = "170:1000")#works
ncbi_byname(taxa="Sardinops melanostictus", gene = c("12s"), seqrange = "170:2000")#fails

[sckott]

👋🏽 previous maintainer here. this pkg is no longer maintained. you can still use it of course

[kdlafferty]

thanks. Will use Rentrez instead.

[maelle]

This repository is about to be archived.

[dlebauer]

👋🏻 new maintainer here: package is now unarchived

[dlebauer]

This error still appears to occur in all of the examples above.

Is this worth resolving? e.g. do the ncbi_* functions still provide a value, or does the rentrez package provide the needed functionality, such that the ncbi_* functions in this package could be deprecated?

[boopsboops]

Is this worth resolving? e.g. do the ncbi_* functions still provide a value, or does the rentrez package provide the needed functionality, such that the ncbi_* functions in this package could be deprecated?

For what it's worth, I use ncbi_byid() function a lot because it returns a nice table. In fact I just noticed this thread when I came here to ask about how to best to fix the http version in the function because it now gives intermittent "Error in the HTTP2 framing layer" errors.

Might be related to OP's problem? Is it worth posting an issue about this?

[dlebauer]

Sorry this took so long. This should be resolved in PR #132

To test it out while the PR is still open:

install.packages("remotes")
remotes::install_github("ropensci/traits@129_ncbi_byname_error") 

If someone checks that this works, that would be much appreciated.

And double points if you are able to provide feedback on the PR!

[dlebauer]

@boopsboops if you found an error with ncbi_byid to report, please do! It would be good to have it documented.