Human-mouse-rat alignments

Glenn Tesler
March 16, 2004

• Assembly versions
• Downloadable data
• Alignment coordinates from PatternHunter
• Anchor coordinates from GRIMM-Anchors
• Synteny block coordinates from GRIMM-Synteny
• Discussion of X chromosome reconstruction (separate document)

G. Bourque, P. Pevzner, G. Tesler, Reconstructing the Genomic Architecture of Ancestral Mammals: Lessons from Human, Mouse, and Rat Genomes. Genome Research, in press, 2004.

B. Ma, J. Tromp, M. Li, PatternHunter: Faster And More Sensitive Homology Search. Bioinformatics. 18(3):440-445, 2002.

Assembly versions

Downloadable data

PatternHunter alignments

We are not distributing the raw alignments. I processed them into coordinate files, which are described below.

The assemblies were first masked at UCSC with RepeatMasker and TandemRepeatFinder. Bin Ma ran PatternHunter to align the non-masked portions of the sequences. The coordinates of the regions that were masked are available at UCSC within these directories:

PatternHunter does base-by-base alignments of 2 sequences. It was run with each combination of two species, and also with each species vs. itself. (I did some processing of the latter for the purpose of analyzing repeats, but we will not be using it in the present set of papers.)

I parsed out the coordinates from the PatternHunter output and put them into my own format. Even the coordinate files without the detailed alignments are still very large.

If you have used PatternHunter directly, please note that the format below is my own format, useful to me for GRIMM-Synteny, and is different than PatternHunter's.

Each line has these fields:

	ID chr1 start1 end1 chr2 start2 end2 sign score evalue

for example:

	1 1 28574062 28575652 X 54629503 54631092 - 1367 0.0
	2 1 28574062 28575652 X 54726654 54728243 + 1355 0.0
	3 1 243490236 243492219 X 15479260 15481161 - 1017 0.0

ID

Due to concatenating a large number of separate PatternHunter output files and then filtering them, PatternHunter's ID number isn't unique or useful any more.

genome number

human-mouse: human=1, mouse=2
human-rat: human=1, rat=2
mouse-rat: mouse=1, rat=2
For example, in the file human-mouse.gz, (chr1,start1,end1) refer to human and (chr2,start2,end2) refer to mouse.

chr1, chr2

The chromosomes are denoted by their name as a string.

start1, end1, start2, end2

The start/end are 0-based on the + strand. The intervals are half-closed, half-open: [start,end).

sign

The sign is '+' or '-'. Negative means to take the string of bases from the interval as indicated on the + strands of both genomes/chromosomes, but then reverse complement one of those strings to align it to the other. See more on this below.

score

Higher scores are better alignments.

evalue

Lower evalues are better alignments.

Note on signs

   position   0  1  2  3  4  5  6  7  8  9
   + strand  [A] C  T  A  G  C  C  A  T  G
   - strand  [T] G  A  T  C  G  G  T  A  C

Just to clarify why I'm describing that if it seems like the only way to you, I've also seen this:

   position   0  1  2  3  4  5  6  7  8  9
   + strand  [A] C  T  A  G  C  C  A  T  G
   - strand  [T] G  A  T  C  G  G  T  A  C
   position   9  8  7  6  5  4  3  2  1  0

where that position is identified as 0 on the + strand and 9 on the - strand. I've also seen variations where one or both strands are 1-based, the strands are indicated by positive/negative coordinates, etc.

GRIMM-Anchors coordinate files

1. in all other species they are consecutive in the exact same order as the first species or the exact reverse order (with all signs inverted as well)
2. and the anchors are close in all species (sum over species of gap between two consecutive anchors is < 50000 in 2-species comparisons or < 75000 in 3-species comparisons).

In the 2-species version, each line has these fields:

	ID chr1 start1 end1 chr2 start2 end2 sign

	ID chr1 start1 end1 sign1 chr2 start2 end2 sign2 chr3 start3 end3 sign3

ID

Anchors are sorted in order by first genome and numbered consecutively.

genome number

human-mouse: human=1, mouse=2
human-rat: human=1, rat=2
mouse-rat: mouse=1, rat=2
human-mouse-rat: human=1, mouse=2, rat=3

chr1, chr2, chr2

The chromosomes are denoted by their name as a string.

start1, end1, start2, end2, start3, end3

The start/end are 0-based on the + strand. The intervals are half-closed, half-open: [start,end).

sign (2 species)

The sign is '+' or '-'. Negative means to take the string of bases from the interval as indicated on the + strands of both genomes/chromosomes, but then reverse complement one of those strings to align it to the other. See more on this above.

sign1, sign2, sign3 (3 species)

Each sign is '+' or '-'. Positive means to take the string of bases from the interval as indicated on the + strand of the appropriate genome. Negative means to take from the positive positions for that genome and then reverse complement the string. See more on this above.

GRIMM-Synteny coordinate files

The file blocks_hmr_1000000 is human-mouse-rat blocks at 1 Mb resolution. The genome #'s in this file can be derived from the filename (order h,m,r: genome 1 = human, genome 2 = mouse, genome 3 = rat). Similarly, in hr_1000000, genome 1 = human and genome 2 = rat. And so on.

Resolution x (300000 or 1000000) means that the length of the every block in the first genome is >= x and the GRIMM-Synteny "gap threshold" is x (in the 2-species files) or 1.5*x (in the 3-species files). We may change the definitions and parameters in future runs.

The first line of the file is "# " plus tab-separated names of the columns. Each subsequent line is a 2 or 3-way synteny block. The files are sorted in order by chromosome and then starting position in genome 1.

cluster_number

The ID assigned to that synteny block.

org1.chrom, org1.start, org1.end, org1.sign

are the chromosome # (as described above), start and end (as closed interval), and sign (1 or -1), for whatever genome 1 is in that file.

org1.pos

an integer sequence 1,2,... giving the order of the synteny blocks on chromosome org1.chrom of genome 1.

org2.*, org3.*

similar