A serious bug affecting representative subsets of original-style
sequences in ASTRAL 1.57 and 1.59 has now been fixed.  All original-
style subsets filtered by E-value and percent identity contained
extra domains.  Genetic domain subsets, the default sets in
1.57 and 1.59, were not affected by the bug.  The affected subsets
have been regenerated correctly, and are now available on our
website through the normal download mechanisms.  Files containing
the subsets which were originally released are still available
through direct access to ASTRAL files, or by downloading a tarball
of all files; however, these files have been clearly identified
by prepending the word "broken" to their names.  Thanks to John
Spalding for reporting this bug.  (Fixed 9/10/02)

The sequences based on ATOM records for four domains are truncated
in ASTRAL 1.59.  The entries for these domains in the RAF file were
corrupted, leading to the bug.  The RAF file has been corrected, and
the ATOM sequences will be fixed in the next release.  These are
the four domains affected:
d1hcja_
d1hcjb_
d1hcjc_
d1hcjd_

There are some bugs in the underlying PDB files which we do not attempt
to fix; we have identified files with missing residues in the SEQRES
records (which were present in the ATOM records), and cases where the
residues in the SEQRES and ATOM fields do not match.  These are lists
of the domains affected:

Domains with missing SEQRES records (cases where coordinates for a
residue are given in the ATOM records, but the residue is not included
in the SEQRES records):
d1fcs__
d1ifa__
d1mdc__
d9pti__

Domains with mismatches between SEQRES and ATOM records (cases where
a coordinates for a residue are given in the ATOM records, but the
SEQRES records indicate a different residue should appear at that position.
d2fam__ k  146  -> a
d4ccp__ d  272  -> n
d1fna__ r    6  -> g
d2plv1_ l    7  -> s
d1s01__ p   56  -> n
d2bce__ n  187  -> q
d4icd__ e  331  -> q
d8gpb__ k  840  -> a
d3gbp__ q   25  -> e
d1pda_2 a  241  -> g
d2pf2_2 r   15  -> e
d1hra__ p    1  -> m