Changes and additions from ASTRAL 1.63 to 1.65:


* Server Change

The ASTRAL web server has been moved from http://astral.stanford.edu/
to http://astral.berkeley.edu/.  Please update your links.


* ASTEROIDS

This is a continuously updated version of ASTRAL, described in the
2004 NAR paper.  We have changed the algorithm used to generate
ASTEROIDS slightly from that described in the paper:

1) Alignments and hidden Markov models (HMMs) are generated for each
   ASTRAL family, as well as for each superfamily, using the same
   method described in the paper.

2) During weekly classification of new PDB entries, HMMs from both the
   family and superfamily libraries are used to predict domains in the
   sequences.  In addition, sequences of the new PDB entries are
   compared to all existing ASTRAL sequences using BLAST.  A region of
   PDB sequence is annotated as belonging to the same ASTRAL
   superfamily as a matching ASTRAL sequence or HMM if the E-value
   resulting from either HMMER or BLAST is more significant than
   10^-4.  Overlaps of up to 10 residues between multiple HMMs are
   allowed, and regions of sequence matching several domains are
   assigned to the one with the more significant E-value (from either
   method).  Longer overlaps result in the entire domain prediction
   being rejected, and flagged for manual review in cases where two
   different superfamilies are predicted to overlap.

3) As described in the paper, remaining unclassified regions are
   annotated using Pfam, and remaining regions of 20 or more residues
   in length are annotated as potential unclassified domains.  (The
   latter limit was originally set at 50 residues in the paper, but
   changed to 20 to be consistent with the rest of ASTRAL).

4) Pfam release 10 was used to create ASTEROIDS in ASTRAL 1.65.

5) The FASTA headers for ASTEROID sequences indicate the source of
   each prediction as well as the E-value of the prediction.

   Examples:

   >u1bjya2 (a:67-207) ASTEROIDS sf:[ASTRAL-1.65-SF]a.121.1 logE:-88.89 Tetracycline Repressor

   This sequence is the second predicted domain from the PDB entry
   1BJY chain A, and includes residues 67-207.  It is predicted to
   belong to the SCOP superfamily a.121.1 on the basis of a HMMER hit
   against the corresponding SCOP superfamily HMM, with an E-value of
   10^-88.89.

   >u1eaoa_ (a:17-129) ASTEROIDS sf:[ASTRAL-1.65-Fam-b.2.5.6]b.2.5 logE:-101.25 Runt-Related Transcription Factor 1

   This sequence is the only predicted domain from the PDB entry 1EAO
   chain A, and includes residues 17-129.  It is predicted to belong
   to the SCOP superfamily b.2.5 on the basis of a HMMER hit against
   the SCOP family HMM for the family b.2.5.6, with an E-value of
   10^-101.25.

   >u1a6ea1 (a:17-145) ASTEROIDS sf:[ASTRAL-1.65-BLAST-d1a6da1]a.129.1 logE:-59.70 Thermosome

   This sequence is the first predicted domain from the PDB entry 1A6E
   chain A, and includes residues 17-145.  It is predicted to belong
   to the SCOP superfamily a.129.1 on the basis of a BLAST hit against
   the SCOP domain d1a6da1 with an E-value of 10^-59.70.

   >u1a6ea5 (a:404-545) ASTEROIDS sf:[UNMATCHED] logE:UNMATCHED Thermosome

   This sequence is the 5th predicted domain in the PDB entry 1A6E
   chain A, and includes residues 404-545.  It's a long region for
   which no significant (E-values of 10^-4 or better) were found by
   BLAST or HMMER against current SCOP sequences, families, or
   superfamilies.


* PDB Chain Sequences

In versions of ASTRAL prior to this one, residues with no atomic
coordinates at the beginning and ends of the sequences were removed,
as is the case with SCOP domain sequences.  Beginning with this
version of ASTRAL, we include these residues in the PDB chain
sequences.


* PDB-Style File Improvements:

We now generate PDB-style files for domains in SCOP classes 8 and
above, unless the domain contains no amino acid residues of known
type.  (This is the case for several low resolution or polynucleotide
entries.)  When downloading files, the set of files for which the data
has changed between releases can now be downloaded separately, to save
on bandwidth.


* Bug Fixes:

More manual edits have been made to the RAF file, so some sequences
have been corrected.