|2.58 ||08/04/2014 ||
- 'Select domain' function, Help text links, narrow warning/help text pane and 'Arrange by SCOP familyi' fixed.
|2.56 ||12/11/2013 ||
- Check all button on match results.
|2.55 ||26/3/2013 ||
- Multiple alignment results JMol options.
|2.54 ||22/2/2013 ||
- Winter Update.
- Fix the View link on 'Submission Page'
|2.51 ||2/4/2012 ||
- Problem with one or or more processes not completing, when specifying low persentage match.
- Results page headers are now help links.
- Fixes for Google Chrome.
|2.50 ||14/12/2011 ||
- Problem with one or two CPUs slow to complete fixed.
- PDB codes in match list are links.
|2.49 ||15/11/2011 ||
- Exclusions list added to weeekly perocessing capability.
|2.47 ||14/11/2011 ||
- Web page links made relative so that hppts requests are handled correctly.
|2.47 ||12/11/2011 ||
- RMSD values correctly displayed for multi-alignments. (Some values were zero).
- Alignment number displayed (Alignment i of n) in alignment details page.
|2.46 ||26/10/2011 ||
- Fixed cut-and-paste problems in HTML pages.
- Summer make over. 64bit native code and data files.
- Fixed a bug preventing Jmol applet from starting in
certain MS Windows setups.
- A major revision of PDBeFold database structure due to the
increased size of PDB archive. No changes in results
and functionality with the except for recent PDB entries
now appearing in database searches
- A facility to reject unfair-use users has been added.
- New checkbox "enforce sequence identity" is added to the
Submission Form when both query and target are given
by single structures (PDB/SCOP entries or coordinate files).
The checkbox is unchecked by default, which corresponds
to the way PDBeFold used to work in earlier versions. When the
checkbox is checked, PDBeFold tries to calculate initial
superposition from sequence alignment, rather than from
matching secondary structure graphs. This option should be
used when a sequence-identical alignment is required, even
if it results in a lower geometric score. A typical example
is the calculation of transformation matrix between identical
chains in ASU, not related by crystal symmetry (NCS matrix).
- New checkbox "superpose whole entries" is added to the
result pages. When it is unchecked, only selected
parts of entries (chain, range, domain) are superposed
and visualised/downloaded. When the checkbox is checked,
PDBeFold applies superposition matrices to the whole entries
or coordinate files and sends them for visualisation
- Data sent to rasmol/jmol are now uniquely named, which
overcomes cache problems caused by custom browser
and proxy settings
- Wait pages are renamed at every reload, also for
forcing browsers and proxy servers to obtain the
- C-alpha alignment added as an option to XML output
in off-line mode.
- Fixed yet another bug in multiple alignment module,
which caused crash under certain conditions.
- Fixed bug which caused occasional crashes in XML
output of C-alpha alignment.
- Chain IDs in capital and small letters are now
differentiated (i.e. chain "A" is different of "a").
- Fixed few bugs which caused PDBeFold to crash when downloading
results (XML and sequences) at certain conditions.
- A bug causing termination of computational jobs
in Multiple Alignment has been identified and
fixed. Due to this bug, PDBeFold was showing never
ending calculations for some rare type of
- Visualisation in Jmol added.
- Original residue sequence numbers and insertion codes
are now preserved in coordinate data sent for
visualisation in both Jmol and Rasmol. Wherever the
opposite is written in documentation should be ignored
now. The chain ID may differ in the visualisation
scripts, where originally duplicated. Atom serial
numbers are used for selection and they do differ
- Fixed bug in parsing lists of PDB and SCOP entries
in Submission Form for multiple alignment.
- Fixed a bug, which in rare instances prevented PDBeFold from
loading the results of multiple alignments, so that session
looked like never finishing calculations
- File parser was slightly modified such as to accept some
roque formats like Windows text UTF-8
- Session ID changed to encrypted form
- PDBeFold may work with gzipped or compressed PDB archives.
This affects only details of local installations and
makes no difference to user interface or results.
- PDBeFold accepts compressed (*.Z) files for upload, those
may be coordinate files, tarballs, code lists etc.
Any file to be uploaded to PDBeFold may be gzipped (*.gz)
or compressed for faster transmission.
- Introduced a new option of uploading a list of
PDB or SCOP codes for Multiple Alignment.
- Improved diagnostics on inacceptable structures
in Multiple Alignment submissions.
- The sharing mechanism of accessing the PDBeFold log file
has been improved, which should prevent log crashes at
multiple simultaneous access in busy environments.
The fix has no effect on PDBeFold's calculation results.
- Fixed bug that caused a permanent PDBeFold misconfiguration in
mixed-platform setups (e.g. Sun+Linux) under certain
- Fixed bug that prevented submission of structures for
multiple alignment from the list of (pairwise) results
after resorting the latter.
- Fixed bug that caused the output of incorrect sample of
xml input file used by off-line (command-prompt) PDBeFold.
This also affects the copy of input data in the output
- Fixed bug that prevented PDBeFold from taking entries without
sidechains but with complete mainchain.
- XML tag <3DAlignment>, which was in violation
of XML naming conventions, changed for
<StructAlignment>. The tag was used in PDBeFold's
output XML (downloadable) files for pairwise structure
- Prevented occasional server failures due to NFS delays at
- Improved server's diagnostics in case of submitting
non-aminoacid chains or chains made of only backbone C-alphas.
- Fixed a bug blocking submission of SCOP matches for
Multiple Alignment from the result list.
- PDBeFold PDB parser is softened to allow for minor
violations of PDB specifications in coordinate files
like disagreement in atom names between ATOM and
ANISOU etc. records.
- Fixed bug in displaying the results of multiple
alignment, which caused indefinite looping in
- Fixed bug in generating Rasmol scripts for individual
structures in multiple alignment result page, which
caused PDBeFold to crash erratically on pushing the
- Algorithm of multiple 3D alignment improved in part
concerning the analysis of multiple atom contacts for
candidate mappings. In many instances, this results
now in longer alignments with higher Q-score.
- PDBeFold database of precompiled structures is aggressively
optimized, which resulted in 30-50% gain in space
and access time. PDBeFold should server queries in general
faster now, with most effect seen for setups on the
edge of hardware requirements.
- Fixed bug, which, in the case of a residue range
selection from query/target structures, caused
sending the whole structure in the PDBeFold rasmol
scripts instead of sending only the selected part.
- Enhanced security of PDBeFold CGI web pages.
- Fixed bug making specification of PDB and SCOP codes
in PDBeFold URL queries case-sensitive.
- Introduced suggestion of SSE similarity thresholds
in case when querying an archive does not yield
- Number of matches obtainable for each pair of structures
with option "Best Matches Only" switched off is
now limited such as to avoid enormous number of
alternatives in some cases (because of which PDBeFold
could hang for a long time).
- Fixed bug in reading mmCIF coordinate files because of which
Ca atoms were not
recognized under certain conditions.
- Added facilities for working with multiple alignments through
the URL link and off-line, see
- Fixed bug in generating the Rasmol scripts, because
of which the colouring of superposed structures might
be incorrect in rare instances.
- Fixed bug causing occasional misinterpretation of calcium
and Ca atoms in PDBeFold
database under certain conditions.
- Fixed bug, due to which the correspondence between
SCOP domains and PDB entries was being established
on the level of entries, rather than chains. Now
different chains of a PDB entry are, if that should
be so, put in correspondence with different SCOP
- Fixed bug, due to which the secondary structure assignment
in the output table of
Ca was wrong or
missing for SCOP domains starting not from the begining
of a chain.
- Algorithm of multiple alignment in 3D tuned up as to
perform better in situations when common secondary
structure motif contains only a few (1-3) SSEs.
- The search in PDB and SCOP archive may be narrowed by
uploading the list of PDB or SCOP entries which should
be taken as target structures.
- Sub-structures may be formed by selecting a range of
residues (in addition to chain and domain selections).
- On-line documentation (help) updated.
- Added incremental procedure of updating the PDBeFold's precompiled
- Fixed bug in the calculation of P-values, which caused
an appearance of infinite P-values in rare instances.
- When connectivity check is turned off and
in parts of chains being misconnected, the match is selected
by italic in the list of results.
- When a structure is matched against PDB or SCOP archives,
the results may be arranged into groups by SCOP families.
The sorting order is preserved within the groups and between
the groups, based on the sorting score of the group's leading
hit. The list of results may be paged up and down by SCOP
groups, and brought to a SCOP group directly. See new control
elements following the list of results.
- Fixed bug in the module decoding query string (when PDBeFold is
see linking PDBeFold).
- When PDBeFold is linked through a URL, local files may be specified
as query/target structures (cf.
- Added submission of tarballs of structures for
- Fixed bug in the representation of multiple alignment
of secondary structure elements.
- Added selection of matches. Selected matches may be specified
for downloading target sequences and alignments, brought on
top of the results and submitted for multiple alignment.
- Added XML and plain text output of results of multiple
- Removed bug which excluded chains without SSEs in case
when many-chain structures were compared as a whole
rather then chain-by-chain.
- Improved synchronization of PDBeFold's parallel computational
jobs such as to minimize the effect of file system caching
on CPU clusters. The results should now load faster and
should not get lost due to PDBeFold desynchronization.
- On-line Help updated as to cover the multiple alignement.
- PDBeFold is extended for multiple 3D alignment. More than
2 structures may be aligned simultaneously, the results
are scored in respect to coordinate consensus and each
aligned structure. Both SSE and Ca
alignments are given in the output. The pairwise/multiple
alignments are chosen on the very top of the PDBeFold
- Linking the results to PDBe Atlas Pages is added.
- A bug is removed, which caused premature end of
parsing PDB files containing duplicate sequence numbers,
empty lines and syntactical errors in non-coordinate
section, in data uploaded as a
file set tarball.
- A bug is removed, which prevented PDBeFold from properly
selecting chains with numerical identifiers, when
those are specified in the Submission Form.
- PDBeFold parser is further modified to properly read PDB/mmCIF
files coming from different platforms and therefore having
different line terminators. As practice shows, in many
instances uploaded PDB-like files have empty lines, which
is against PDB syntax; such lines are now silently ignored.
- Result pages for individual matches now display
secondary structure alignment in terms of strings
composed of "H" for helices and "S" for strands, so
that match topology may be seen immediately.
- Result pages for individual matches now display
a direct reference to SCOP families where
- SCOP domain may be specified as a query structure
and as a single target structure. Both SCOP domain name and
sunid may be used for identifying the structure.
- Fixed bug which caused spliting SCOP domains consisting
of more then 1 chain, if matching individual chains
is set on.
- A bug, which enforced ordering of SSEs found in different
chains, is removed. This affected matches of structures or
domains consisting of more than one protein chains.
- In structure-to-structure regime, selection of chains
or domains is now allowed also for the target structure.
- Internal threshold value are slightly adjusted according
to experience accumulated. This does not invalidate any
previous results, however new hits may emerge in some
- Coordinate files with <CR> line terminators (Windows
standard), as well as <LF><CR> (MS DOS) are now
Q-score is added to the PDBeFold output.
- Q-score is made default sort subject (instead of RMSD).
- Sequence identity is reported in result pages and
in downloaded files.
- Aligned sequences may be download in FASTA format.
- PDBeFold on-line help updated in order to reflect last changes.
- 3D Ca alignment algorithm
is modified to better handle the situations when
only one SSE has been matched (extremely low structural
- Tune-up of parameters controlling the 3D
Ca alignment for improving
the ratio of alignment length to RMSD.
- Modified convergence conditions for 3D
Ca alignment iterations.
No change in results was noticed, however PDBeFold
is 20% to 30% faster.
- Advanced algorithm for 3D Ca
alignment, which includes a modified target function.
You might see a slight change in the results for
remote structural neighbours, which should be seen
in longer alignments and/or lower RMSDs.
- Changed the mechanism of PDBeFold session expiration. Now
the session expires after 4 hours since the last access
to its results. If the results were not accessed
immediately after the calculations (e.g. if the wait
(progress) page was bookmarked and browser switched to
another site or page), the session expires in 48 hours
after the calculations have actually finished.
- Results for individual matches, including the
3D Ca alignments, may now
be downloaded from the result page in either XML or
plain text format.
- PDBeFold uploads the query and target structures into separate
session sub-directories, which eliminates a possibility
of file name conflict with PDBeFold own files.
- PDBeFold accepts
lists of pairs for multiple pairwise matching of
structures uploaded in a
file set tarball (a special
query option is added).
- PDBeFold accepts gzipped coordinate (PDB and mmCIF) files
- PDBeFold accepts
file sets as a
target (a special target option is added).
- The transformation matrix now superposes the target structure
over the query (it was the other way round before)
- PDBeFold version is added into output XML file
- WWW linking is added, so that
PDBeFold queries may be launched from any web site