2.0.0.rc3

• On sub-URI deployments, back-button would result in no databases to be displayed in the search form (#462). Fixed that. Thanks to Xiang Zhang for reporting the issue and testing the fix.

2.0.0.rc2

1. A programming error prevented databases from being deselected when you go back to the search form after submitting a query. That's now fixed. Thanks to Norman Johnson for reporting the issue.
2. Database creation step could skip FASTA files containing too many Ns at the beginning of the file. Fixed that. Thanks to Michał T. Lorenc for reporting the issue.
3. During the initial setup, SequenceServer would print the message 'config file not found' multiple times. Fixed that.
4. Results page would fail to load completely if the submitted queries resulted in no hits. Fixed that.
5. Give users an option to sign up for release and other important announcements during the initial setup.
6. Highlight the use of non-default parameters by changing the background of advanced parameters field to yellow when filled.
7. Include SequenceServer version and a link to our paper in the search summary at the top of the results page.

1.0.13

• Fix BLAST version comparison to check compatibility, so that 1.0 series can work with BLAST 2.10.0 (issue raised by Takekazu KUNIEDA)
• Point to our MBE paper instead of biorxiv when SequenceServer is terminated

2.0.0.rc1

Summary of changes since version 1.0:

1. Support for BLAST 2.10.0 and the new BLAST database format.
2. Results now open in a new page. You can share the link with colleagues or
   bookmark the page and return to it at a later date. Results are saved for 30 days.
3. We have integrated three new visualisations that facilitate interpretation of BLAST results:
   Length distribution (histogram) of all hits of a query. This is inspired from our work with GeneValidator (https://academic.oup.com/bioinformatics/article/32/10/1559/1742817) and can, for example, help quality-assessment of gene predictions.
   Kablammo visualisation (https://academic.oup.com/bioinformatics/article/31/8/1305/212772) for each query-hit pair indicates clearly which parts of the query match which part of the hits.
   Circos-style plot of all queries and their top hits. This is similar to the circoletto tool (https://academic.oup.com/bioinformatics/article/26/20/2620/194655) but for a small number of queries. For example, it can help visually testing for conserved synteny or gene duplication.
4. We provide a command-line mechanism for importing a BLAST XML file and visualising it in SequenceServer. This works for DIAMOND as well. We do not yet provide a graphical user interface for this.
5. The 30 hit limit for FASTA download has been removed. In addition, you can now download
   raw pairwise alignments.
6. Better support for long-running BLAST searches, for rendering large BLAST results, for
   integrating as part of other websites.
7. Improved error handling, security enhancements, and a new look.

Changes since the last beta release:

1. Upgraded to BLAST 2.10.0
2. We have improved the ability to render "large BLAST results" (thousands of hits). We expect
   this to facilitate visualisation of large gene and repeat families.
3. Form data is now restored when you press back-button in the browser from the results page.
4. We have a logo and a new look!

2.0.0.beta4

New features since 1.1.0-beta12

• Update to BLAST+ 2.9.0
• Ability to import BLAST results in XML format from DIAMOND and BLAST versions 2.2.30 - 2.9.0
• Ability to pre-populate textarea with server-sent query sequences (thanks to Richard Challis for feature suggestion and initial implementation)
• On startup, sequenceserver now tries to determine the ip and the hostname of the server and prints two URLs that may be suitable for sharing with colleagues
• Recognise PFAM and RFAM hit/subject ids and include a link to the hits PFAM or RFAM page (credit: Tomas Pluskal)

Bug fixes and refinements

• In a suburi deployment (e.g., antgenomes.org/blast), take suburi into account when constructing url for the results page (thanks to Guy Leonard for reporting the issue)
• Use more specific regex for determining non -parse_seqids databases. Otherwise, presence of 'gnl|' in sequence ids would disable sequence retrieval (thanks to Josh Goodman for reporting the issue)
• Length of query and hit sequences are now easily accessible in the report page. This is included in the query and hit header, on the right hand side, in a gray font
• Restored the ability to fold and un-fold individual hits (credit: Josh Goodman)
• Removed .00 from decimal numbers (i.e., 100.00% is now shown as 100%)
• Fixed vertical alignment of links and buttons provided for each hit
• Removed trailing comma in HSP statistics line for BLASTP
• Properly wrap text in the summary overview on top

Circos visualisation:

• Sequence lengths were erroneously downscaled by 100. Fixed that
• Formatting of e-value and identities in the tooltips was inconsistent with the rest of the report. Fixed that
• Ribbons had no colour for the specific case of 1 query and 2 hits. Fixed that
• Filter out queries with no hits before drawing the visualisation
• Change to more meaningful title 'Chord diagram: queries and their top hits’

Length distribution of all hits for each query:

• When query and database are in different sequence space (DNA or protein), normalise query length to match database space (#409)
• Reduced the number of ticks in the y-axis otherwise the ticks looked too “squeezed up” when there were too many hits
• Round up the number of ticks on x-axis so that the scale ends consistently in a major or a minor tick

Code quality improvements

• Fixed many of the warning messages generated by React in browser console
• Fixed style guide violations in both Ruby and JavaScript code
• Expanded test coverage and improved automated testing

1.0.12

Point to our new MBE paper in the footer.

1.1.0.beta12

Changelog

• Write default BLAST options that were previously defined only in the code
  to config file during setup. Since old config files won't have this info
  available, retain the old behaviour of automatically adding '-task blastn'
  to BLASTN searches and log this during startup.
  (issue raised by Niek Art)
• Multipart database volumes can have 3 digit suffixes (e.g.
  refseq_genomic.100). Our regex allowed for only 2 digit
  suffixes (e.g. nr.00). Fixed that.
  (issue raised by Niek Art)
• Show location of config file by default during startup
• Code-style improvements.
  (credit: Emeline Favreau, Esteban Gomez)

1.1.0.beta11

• Display of large results are handled by rendering results of fifty queries at
  a time. However, in each cycle of update, instead of focusing only on the
  next fifty queries, React would automatically also check if previously
  rendered queries needed to be updated as well. This resulted in each cycle of
  update to take more and more time. Prevent that by returning false from
  shouldComponentUpdate lifecycle method if the query is already rendered.
• Kablammo / graphical HSP overview:
  • Reverse order of HSPs for drawing so that the HSP with stronger evalue
    are drawn last, and thus end up on top of HSP with weaker evalue if
    overlapping.
  • Handle hover events more efficiently.
• Copy-pasting hit sequence from sequence viewer should eliminate whitespace
  introduced by grouping of residues, but it didn't. So disabled grouping of
  residues in sequence viewer so that the sequence can be copy-pasted as it
  is.
• SVG and PNG download links would break into two lines on smaller windows.
  Fixed that.
• Do not change padding when hit is selected. Further, apply orange border
  to selected hits only on the left side instead of on all sides.
• Downloading alignments of selected hits wouldn’t work if sequence ids
  included a pipe character. Fixed this.
• Last line of the alignment would sometimes not be shown. Fixed that.
• Expanded error handling:
  • Inform user if BLAST's XML output is invalid, and suggest deduplicating
    sequence ids or rerunning with a single database (issue #194) (credits:
    Ben Woodcroft, Tomáš Pluskal, and others).
  • Inform user if BLAST ran out of memory and ask to retry with smaller
    query. Should be rare, but is good to know when this happens.
  • Inform user if SequenceServer doesn't have permission to create job
    directory. Should be particularly helpful when setting up with
    Apache.
  • Inform user if SequenceServer/BLAST runs out of disk space.
  • Inform user if BLAST produced empty output.
• Update Dockerfile so that docker image is built from the source. This
  ensures that the tag applied to docker image and software version are
  in sync. (credit: Michał T. Lorenc)
• Don't use symlinks inside of the repository as symlinks are not
  supported on all platform and can thus cause installation to
  fail.
• Ensure the new version can be deployed to a suburi.

1.1.0.beta10

Features

• Added a button to select all databases.
• Meta data on top of report page now includes date (as well as labels
  for program, databases, parameters).
• Reduce overall amount of vertical space on the report page so that more
  results fit in the same height.
• Detailed space reduction effort were needed for pairwise alignments,
  which now use almost 2x less vertical space. Importantly, alignments are
  now "pretty formatted" in the browser instead of on the server so that all
  available horizontal space is known beforehand. Previously we were restricted
  to assuming 60 characters per line.
• For displaying pairwise alignments we previously used nested tables. We now
  use list of divs and pre tags. This reduces memory usage and increases speed
  (crucial for large reports).
• Eliminate borders from around graphs to reduce clutter and free up vertical
  space.
• Add % sign for when identity, gaps, positive are shown as percentage.
• Use more meaningful headers for hits and hsp overview graphs.

Bug fixes

• Consecutive blank spaces in the middle line of the alignment were truncated
  while parsing BLAST’s XML output. Fix it.

1.1.0.beta8

• Fix programming errors in Report.fetchResults.
• Change title of hits overview to matching sequences.