This file is from:
https://hgdownload.soe.ucsc.edu/goldenPath/hg38/cactus447way/README.txt
This directory contains a Cactus multiple alignment of 447 mammalian genomes,
including 233 primates, to the human genome (hg38/GRCh38, Dec. 2013). The
alignment is single-coverage: one aligned row per species in each block.
-----------------------------------------------------------------------------
CORRECTED ALIGNMENT (2026 rebuild)
These files were rebuilt in 2026 to correct a known bug in the Cactus/hal2maf
"size" field (ComparativeGenomicsToolkit/cactus issue #1201), in which the size
value on some alignment "s" lines did not match the number of aligned bases in
the block. That bug was present in the previous (2023) files served here.
The corrected data is the alignment authors' regenerated single-coverage MAF
(447-mammalian-2022v1.fix2.single, published at https://cglgenomics.ucsc.edu/ ,
corrected November 2025). We split it per chromosome, renamed the reference rows
to UCSC chromosome names, and added the same gap/break (iRows) annotations as
before.
Because the corrected alignment was regenerated by the authors rather than
merely patched, it does not correspond block-for-block to the 2023 files. If you
compare with the previous downloads you will see a different block structure and
a higher covered-base total:
2023 files: 108,226,203 alignment blocks, 3,023,308,066 bases covered
2026 files: 97,277,175 alignment blocks, 3,099,750,718 bases covered
The 76.4 Mb rise in covered bases does not mean more species aligned. The
underlying Cactus alignment was built with an older version of Cactus that
placed little penalty on aligning to unknown (N) bases, so it aligns across
hg38 assembly gaps. Blocks lying over those gaps were filtered out of the 2023
files but are present in the 2026 files, and they account for much of the
increase: on chrY alone, 4,180 blocks covering 33.5 Mb have an all-N human row,
against 12 blocks covering 40 kb in the 2023 file. Such blocks contain no human
sequence, and the Genome Browser displays nothing at those positions.
Both the 2023 and 2026 files are single-coverage; this is a corrected
regeneration of the same single-coverage alignment, not a change of coverage
model. The aligned sequence is the authors' corrected data.
-----------------------------------------------------------------------------
Assemblies used in this alignment:
Human - Homo sapiens
Dec. 2013 (GRCh38/hg38) (reference)
This alignment was created by making three edits (using Cactus) to the 241-way
mammalian Zoonomia Cactus alignment ( https://cglgenomics.ucsc.edu/data/cactus/ ):
1. One additional cat genome, "Felis_catus_fca126" (GCA_018350175.1), was added
as a sister taxon to the existing "Felis_catus".
2. Five additional canine genomes were added: canFam4, "Canis_lupus_dingo"
(GCA_003254725.1), "Canis_lupus_orion" (GCA_905319855.2),
"Nyctereutes_procyonoides" (GCA_905146905.1) and "Otocyon_megalotis"
(GCA_017311455.1). "Canis_lupus" from the Zoonomia alignment was renamed
"Canis_lupus_VD" to reflect that it is a "village dog", not a "wolf" sample.
3. The 43-species primates clade from the Zoonomia alignment was removed and
replaced with a 243-way primates alignment, adding 200 primate species. The
primate genome assemblies come from the Primate Genome Project:
Kuderna, L.F.K., et al. A global catalog of whole-genome diversity from 233
primate species. Science Vol. 380, No. 6648, 906-913 (2023),
DOI: 10.1126/science.abn7829
Phylogenetic tree
The phylogenetic tree was established by the research described in the paper:
Lukas F. K. Kuderna, et al. A global catalog of whole-genome diversity from
233 primate species. Science Vol. 380, No. 6648, 906-913 (2023),
DOI: 10.1126/science.abn7829
Files in this directory:
- hg38.447way.nh.txt - phylogenetic tree used to guide the Cactus alignment
- hg38.447way.commonNames.nh.txt - same tree with the common names
- hg38.447way.scientificNames.nh.txt - same tree with the scientific names
- cactus447wayFrames.bb (~161 MB) - reading frames for the display of
amino-acid coding regions at base level
- cactus447waySummary.bb (~990 MB) - a bigBed file with a pre-calculated maf
coverage graph for viewing larger regions
- hg38.cactus447way.bb (~1.4 TB) - the alignment in bigMaf format, used by the
genome browser track
- maf/ - the per-chromosome alignments (bgzipped MAF, ~1.4 TB compressed, 194
files), with annotations for gap context and genomic
breaks in the underlying assemblies
- md5sum.txt - MD5 sums to verify downloads of the files in this directory
- maf/md5sum.txt - MD5 sums to verify downloads of the files in maf/
For a description of the multiple alignment format (MAF), see
http://genome.ucsc.edu/goldenPath/help/maf.html .
PhyloP conservation scores for these alignments are available at:
https://hgdownload.soe.ucsc.edu/goldenPath/hg38/phyloP447way
The entire set of data in this directory is approximately 2.7 TB.
To download a large file or multiple files from this directory, we recommend
that you use rsync rather than downloading the files via the website:
rsync -avz --progress \
rsync://hgdownload.soe.ucsc.edu/goldenPath/hg38/cactus447way/ ./
---------------------------------------------------------------
All the files in this directory are freely usable for any purpose. For data use
restrictions regarding the individual genome assemblies, see
http://genome.ucsc.edu/goldenPath/credits.html .
---------------------------------------------------------------
Name Last modified Size Description
Parent Directory -
cactus447wayFrames.bb 2026-08-10 04:57 161M
cactus447waySummary.bb 2026-07-26 08:52 1.0G
hg38.447way.commonNames.nh.txt 2026-01-05 15:32 22K
hg38.447way.nh.txt 2025-08-08 09:54 22K
hg38.447way.scientificNames.nh.txt 2023-10-16 15:48 22K
hg38.cactus447way.bb 2026-07-23 00:08 1.3T
maf/ 2026-08-25 19:53 -
md5sum.txt 2026-08-18 17:44 355