[R] Help with ape - read.GenBank()

[Ben Bolker]

James Meadow <jfmeadow <at> gmail.com> writes:

> 
> Hi,
> 
> I am trying to work with the ape package, and there is one thing I am
> struggling with.  When calling the *read.GenBank()* function, I can get it
> to work with an object created like this:
> 
> *>x <- c("AY395554","AY611035", ...)*
> *>read.GenBank(x)*
> 
> However, I am trying to use the function to fetch several hundred sequences
> at once.  So I have been testing with small files, but cannot get the right
> format.  I tried to enter accession numbers with or without headers as a
> .csv and just as a text file, and I have tried *as.vector*, but I always get
> this error:
> 
> >*x <- read.csv("numbers.csv",head=T)*
> *>attach(x)*
> *>x.vect <- as.vector(t(x)**)*
> *>x.seqs <- read.GenBank(x.vect)*
> 
> Error in FI[i]:LA[i] : NA/NaN argument
> 
> I have been trying this with an identical set of accession numbers, and it
> always works with the c() vector but not with the table, so the content is
> not the problem.  I assume I am just missing a simple data structure detail,
> but I am at a loss.  I have even tried a c() vector of more than a hundred
> accession numbers, and it worked quite well, so the length is not a
> problem.

It's not entirely clear how your CSV file is structured, but I'm supposing
that you have all your accession numbers in a single row (with appropriately
matching headers).  (It would be a little bit more "R-ish" if you had
them all in a column, and would make life a little easier, but let's
work with what you have.)

After you use read.csv(), try str(x)

This should tell you that you have a data frame with a lot of columns,
each of which is a factor with a single element and a single level.

>From there try 

m <- as.matrix(x)

str(m) should tell you you have a 1-row matrix of type 'character'

then 

v <- c(m)

should drop the dimensions and leave you with a character vector
(try str(v) again to be sure)

then

read.GenBank(v)

should work

For future phylogenetic/comparative-methods type questions,
it's probably better to write to the r-sig-phylo mailing list ...