Differences in Bos Taurus reference genomes

[CowGirl]

Hi Everyone, I've got a question about the UMD3.1 reference genome for cattle.

The (I think original?) version from Maryland can be obtained here (1) ftp://ftp.cbcb.umd.edu/pub/data/asse...aurus_UMD_3.1/.

There are, however, other reference .fa files available for UMD3.1, for example, from
(2) ftp://ftp.ncbi.nlm.nih.gov/genomes/Bos_taurus/CHR_25/ or (3) ftp://ftp.cbcb.umd.edu/pub/data/asse...aurus_UMD_3.1/

The problem is, these files are not identical. Just looking at the (unzipped) file size for one chromosome (25 – it’s the smallest), considerable differences can be observed. For example:

(Source) Size FileName
(1) 43,619,264 bytes Chr25.fa
(2) 43,517,209 bytes bt_ref_Bos_taurus_UMD_3.1_chr25(1).fa
(3) 43,517,969 bytes bt_ref_Bos_taurus_UMD_3.1_chr25(2).fa

A quick "diff" in Unix shows that the files really aren't the same.
I haven’t found a reason or an explanation for this. As long as we know which reference was used for alignment, we should be ok. Nevertheless, it’s a little unnerving to see that three different “References” that should technically be the same, aren’t…
Does anyone know why this is?

Cheers,
Chris

[rhinoceros]

Hi Everyone, I've got a question about the UMD3.1 reference genome for cattle.

The (I think original?) version from Maryland can be obtained here (1) ftp://ftp.cbcb.umd.edu/pub/data/asse...aurus_UMD_3.1/.

There are, however, other reference .fa files available for UMD3.1, for example, from
(2) ftp://ftp.ncbi.nlm.nih.gov/genomes/Bos_taurus/CHR_25/ or (3) ftp://ftp.cbcb.umd.edu/pub/data/asse...aurus_UMD_3.1/

The problem is, these files are not identical. Just looking at the (unzipped) file size for one chromosome (25 – it’s the smallest), considerable differences can be observed. For example:

(Source) Size FileName
(1) 43,619,264 bytes Chr25.fa
(2) 43,517,209 bytes bt_ref_Bos_taurus_UMD_3.1_chr25(1).fa
(3) 43,517,969 bytes bt_ref_Bos_taurus_UMD_3.1_chr25(2).fa

A quick "diff" in Unix shows that the files really aren't the same.
I haven’t found a reason or an explanation for this. As long as we know which reference was used for alignment, we should be ok. Nevertheless, it’s a little unnerving to see that three different “References” that should technically be the same, aren’t…
Does anyone know why this is?

Cheers,
Chris

Alternative assemblies due to extensive repeats? Just an idea. Funnily enough I encountered Bos taurus a few days ago while blasting against ncbi's reference_genomic db. I had one contig (250 bp or so) that produced 3 identical perfect hits: 1 to a fungi, 1 to Bos taurus, and 1 (the RL correct hit I assume) to some bacterium. Just goes to show that so called reference genomes still have a long way to go before they're 'perfect'.

[Cofactor_DaveMessina]

Hi Chris,

Welcome to the wonderful world of assembly formatting!

One clear difference between (2) and (3) is the number of pieces each is in:

$ esl-seqstat Chr25.fa
Format: FASTA
Alphabet type: DNA
Number of sequences: 1
Total # residues: 42904170
Smallest: 42904170
Largest: 42904170
Average length: 42904168.0

$ esl-seqstat bt_ref_Bos_taurus_UMD_3.1_chr25.fa
Format: FASTA
Alphabet type: DNA
Number of sequences: 19
Total # residues: 42902370
Smallest: 1016
Largest: 12483507
Average length: 2258019.2


From the UMD README, it looks like they might have needed to reformat to suit NCBI's requirements:

The only differences are in the formatting of the AGP file deposited
with GenBank, which we were required to make as a result of inconsistencies
in the original AGP file.

If you'll notice, the difference in sequence length between the two is exactly 1800 bases. Considering that the NCBI version is in 19 pieces, I'll bet that there are 18 100-base stretches of Ns as internal gaps between scaffolds in the 1-piece UMD version that aren't present in the 19-piece NCBI version.


Best,
Dave