[Bioc-devel] A more flexible GenomeInfoDb::mapSeqlevels(): used supported info but don't break with new organisms/toy examples

[Sonali Arora]

Hi Leo,

This bug is now fixed .
JGI2 was being incorrectly returned previously.
The correct style is JGI2.F and from now on, it will always check only 
for JGI2.F

 > library(GenomeInfoDb)
 > seqlevelsStyle("T")
[1] "JGI2.F"
 > seqlevelsStyle("LGI")
[1] "NCBI"
 > mapSeqlevels('LGI', 'JGI2.F')
[1] "T"
 > packageVersion("GenomeInfoDb")
[1] ‘1.3.4’

Thanks for pointing these out.
Best,
Sonali.


On 10/23/2014 3:35 PM, Leonardo Collado Torres wrote:
> Hi,
>
> I just found a tiny bug in .guessSpeciesStyle(). Basically, if the
> style name has a dot, this function incorrectly returns the name of
> the style.
>
> See below:
>
> ## Lets guess the species and style for 'T'
>> GenomeInfoDb:::.guessSpeciesStyle('T')
> $species
> [1] "Populus trichocarpa"
>
> $style
> [1] "JGI2"
>
> ## Same style with the higher lvl function
>> seqlevelsStyle('T')
> [1] "JGI2"
>
> ## Ok, style 'JGI2'
>
> ## In more dtail, this matches 'LGI' in NCBI style for this organism
>
>> GenomeInfoDb:::.supportedSeqnameMappings()["Populus_trichocarpa"]
> $Populus_trichocarpa
>       NCBI JGI2.F
> 1     LGI      T
> 2    LGII      2
> 3   LGIII      3
> 4    LGIV      4
> 5     LGV      5
> 6    LGVI      6
> 7   LGVII      7
> 8  LGVIII      8
> 9    LGIX      9
> 10    LGX     10
> 11   LGXI     11
> 12  LGXII     12
> 13 LGXIII     13
> 14   LGIV     14
> 15   LGXV     15
> 16   LGVI     16
> 17  LGVII     17
> 18 LGVIII     18
> 19  LGXIX     19
> 20   Pltd   <NA>
>
> ## Guessing 'LGI' with the higher lvl function works
>> seqlevelsStyle('LGI')
> [1] "NCBI"
>
> ## Mapping from 'LGI' in 'NCBI' style to 'JGI2' style doesn't work
>> mapSeqlevels('LGI', 'JGI2')
> Error in mapSeqlevels("LGI", "JGI2") :
>    supplied seqname style "JGI2" is not supported
>
> ## Using the correct 'style' name works.
> ## However a user might have expected it to work with 'JGI2' given
> ## that this was the output from seqlevelsStyle('T')
>> mapSeqlevels('LGI', 'JGI2.F')
> [1] "T"
>
> ## Although a user could find the correct name
>> head(genomeStyles('Populus trichocarpa'), 1)
>    circular auto   sex NCBI JGI2.F
> 1    FALSE TRUE FALSE  LGI      T
>
>> packageVersion('GenomeInfoDb')
> [1] ‘1.3.3’
>
> Cheers,
> Leo
>
>