Skip to content

[Bioc-devel] BSgenome changes

9 messages · Leonard Goldstein, Felix Ernst, Kasper Daniel Hansen +1 more

Original
Leonard Goldstein
# 
Dear Bioc team,

I'm following up on this recent GitHub issue
<https://github.com/ldg21/SGSeq/issues/5>. Please see the issue for more
details and code examples.

It looks like changes in Bioc devel result in two copies of the
mitochondrial chromosome for BSgenome.Hsapiens.UCSC.hg19 -- one named chrM
like in previous package versions (length 16571) and one named chrMT
(length 16569).

When using seqlevelsStyle() to change chromosome names from UCSC to NCBI
format, this results in new behavior -- in the past chrM was simply renamed
MT, now the different sequence chrMT is used. Is this intended?

Leonard
1 day later
Hervé Pagès
# 
Hi Leonard,
On 8/12/20 15:22, Leonard Goldstein via Bioc-devel wrote:
Absolutely intended.

There is a long story behind the unfortunate fate of the mitochondrial 
chromosome in hg19. I'll try to keep it short.

When the UCSC folks released the hg19 browser more than 10 years ago, 
they based it on assembly GRCh37:

   https://www.ncbi.nlm.nih.gov/assembly/GCF_000001405.13

See sequence report for GRCh37:

 
https://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/001/405/GCF_000001405.13_GRCh37/GCF_000001405.13_GRCh37_assembly_report.txt

For some mysterious reason GRCh37 didn't include the mitochondrial 
chromosome so the UCSC folks decided to use mitochondrial sequence 
NC_001807 and called it chrM.

However, UCSC has recently decided to base hg19 on GRCh37.p13 instead of 
GRCh37. A rather surprising move after many years of hg19 being based on 
the latter.

   https://www.ncbi.nlm.nih.gov/assembly/GCF_000001405.25/

See sequence report for GRCh37.p13:

 
https://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/001/405/GCF_000001405.25_GRCh37.p13/GCF_000001405.25_GRCh37.p13_assembly_report.txt

Note that GRCh37.p13 does include the mitochondrial chromosome. It's 
called MT in the official sequence report above and chrMT in hg19.

At the same time the UCSC folks decided to keep chrM so now hg19 
contains 2 mitochondrial sequences: chrM and chrMT. Previously it has 
only one: chrM.

So what you see in BioC devel in BSgenome.Hsapiens.UCSC.hg19 and with
seqlevelsStyle(genome) is only reflecting this. In particular 
seqlevelsStyle(genome) <- "NCBI" now does the following:

   - Rename chrMT -> MT.

   - chrM does NOT get renamed. There is no point in renaming this 
sequence because it has no equivalent in GRCh37.p13.

Hope this helps,

H.

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Felix Ernst
    

    

      
      #
    

  


  

      
Hi Leonard, Hi Herve,

I followed your conversation, since I have noticed the same problem. Thanks, Herve, for the explanation of the recent changes on hg19.

The GRCh37.P13 report states in its last line:

MT	assembled-molecule	MT	Mitochondrion	J01415.2	=	NC_012920.1	non-nuclear	16569	chrM

Since the last name is called "UCSC-style-name", wouldn't that mean that chrM has to be renamed to MT and not chrMT?

Thanks again for the explanation.

Cheers,
Felix

-----Urspr?ngliche Nachricht-----
Von: Bioc-devel <bioc-devel-bounces at r-project.org> Im Auftrag von Herv? Pag?s
Gesendet: Freitag, 14. August 2020 01:08
An: Leonard Goldstein <goldstein.leonard at gene.com>; bioc-devel at r-project.org
Cc: charlotte.soneson at fmi.ch
Betreff: Re: [Bioc-devel] BSgenome changes

Hi Leonard,

        
On 8/12/20 15:22, Leonard Goldstein via Bioc-devel wrote:

            

      
      
      
    

        
Absolutely intended.

There is a long story behind the unfortunate fate of the mitochondrial chromosome in hg19. I'll try to keep it short.

When the UCSC folks released the hg19 browser more than 10 years ago, they based it on assembly GRCh37:

   https://www.ncbi.nlm.nih.gov/assembly/GCF_000001405.13

See sequence report for GRCh37:

 
https://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/001/405/GCF_000001405.13_GRCh37/GCF_000001405.13_GRCh37_assembly_report.txt

For some mysterious reason GRCh37 didn't include the mitochondrial chromosome so the UCSC folks decided to use mitochondrial sequence
NC_001807 and called it chrM.

However, UCSC has recently decided to base hg19 on GRCh37.p13 instead of GRCh37. A rather surprising move after many years of hg19 being based on the latter.

   https://www.ncbi.nlm.nih.gov/assembly/GCF_000001405.25/

See sequence report for GRCh37.p13:

 
https://ftp.ncbi.nlm.nih.gov/genomes/all/GCF/000/001/405/GCF_000001405.25_GRCh37.p13/GCF_000001405.25_GRCh37.p13_assembly_report.txt

Note that GRCh37.p13 does include the mitochondrial chromosome. It's called MT in the official sequence report above and chrMT in hg19.

At the same time the UCSC folks decided to keep chrM so now hg19 contains 2 mitochondrial sequences: chrM and chrMT. Previously it has only one: chrM.

So what you see in BioC devel in BSgenome.Hsapiens.UCSC.hg19 and with
seqlevelsStyle(genome) is only reflecting this. In particular
seqlevelsStyle(genome) <- "NCBI" now does the following:

   - Rename chrMT -> MT.

   - chrM does NOT get renamed. There is no point in renaming this sequence because it has no equivalent in GRCh37.p13.

Hope this helps,

H.

            

      
      
      
    

        

      
      
    

        
--
Herv? Pag?s

Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M1-B514
P.O. Box 19024
Seattle, WA 98109-1024

E-mail: hpages at fredhutch.org
Phone:  (206) 667-5791
Fax:    (206) 667-1319

_______________________________________________
Bioc-devel at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/bioc-devel

  
  


  


      

    

    

      
      

        


  

        

      Hervé Pagès
    

    

      
      #
    

  


  

      
Hi Felix,

        
On 8/13/20 21:43, Felix Ernst wrote:

            

      
      
      
    

        
This is a mistake in the sequence report for GRCh37.p13. GRCh37.p13:MT 
is the same as hg19:chrMT, not hg19:chrM.

hg19:chrM and hg19:chrMT are **not** the same sequences. The former is 
NC_001807 and has length 16571 and the latter is NC_012920.1 and has 
length 16569.

Yes, seqlevelsStyle() is sorting out all this mess for you ;-)

Cheers,
H.

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    
3 days later

    

      
      

        


  

        

      Kasper Daniel Hansen
    

    

      
      #
    

  


  

      
In light of this, could we get a version of GRCh37 with only a single
mitochondrial genome?

        
On Fri, Aug 14, 2020 at 6:17 PM Herv? Pag?s <hpages at fredhutch.org> wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Hervé Pagès
    

    

      
      #
    

  


  

      
On 8/18/20 01:40, Kasper Daniel Hansen wrote:

            

      
      
      
    

        
You mean a BSgenome.Hsapiens.NCBI.GRCh37.p13 package? So it would 
contain the same sequences as BSgenome.Hsapiens.UCSC.hg19 but without 
the hg19:chrM sequence?

Certainly doable but note that by using BSgenome.Hsapiens.UCSC.hg38 you 
stay away from this mess. I'm not sure that adding yet another BSgenome 
package would make the situation less confusing.

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Leonard Goldstein
    

    

      
      #
    

  


  

      
Thanks for the explanation Herv?.

Best wishes

Leonard

        
On Tue, Aug 18, 2020 at 10:06 AM Herv? Pag?s <hpages at fredhutch.org> wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
  


  


      

    

    
1 day later

    

      
      

        


  

        

      Kasper Daniel Hansen
    

    

      
      #
    

  


  

      
Well, the presence of two mitochondrial genomes is to fix a mistake by
UCSC. I can appreciate the importance of representing this mistake when you
build off UCSC. But it strikes me as not actually representing the h37
version of the genome, and it seems to me that we want such a
representation in the project - not everything comes through UCSC. But
perhaps I have not given this sufficient thought, this is just my immediate
reaction.

On Tue, Aug 18, 2020 at 8:18 PM Leonard Goldstein <

        
goldstein.leonard at gene.com> wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Hervé Pagès
    

    

      
      #
    

  


  

      
Kasper,

The tradition so far has been to package all UCSC human genomes since 
hg17. We could also start producing BSgenome packages for other non-UCSC 
Human assemblies. We just need to draw a line somewhere. If there is a 
need for it we can make BSgenome.Hsapiens.NCBI.GRCh37.p13 available, as 
I said earlier. Is this what you are asking for?

H.

        
On 8/20/20 03:23, Kasper Daniel Hansen wrote: