# ===========================================================================
#
#                            PUBLIC DOMAIN NOTICE
#               National Center for Biotechnology Information
#
#  This software/database is a "United States Government Work" under the
#  terms of the United States Copyright Act.  It was written as part of
#  the author's official duties as a United States Government employee and
#  thus cannot be copyrighted.  This software/database is freely available
#  to the public for use. The National Library of Medicine and the U.S.
#  Government have not placed any restriction on its use or reproduction.
#
#  Although all reasonable efforts have been taken to ensure the accuracy
#  and reliability of the software and data, the NLM and the U.S.
#  Government do not and cannot warrant the performance or results that
#  may be obtained by using this software or data. The NLM and the U.S.
#  Government disclaim all warranties, express or implied, including
#  warranties of performance, merchantability or fitness for any particular
#  purpose.
#
#  Please cite the author in any work or product based on this material.
#
# ===========================================================================


The NCBI SRA Software Development Kit
Contact: sra-tools@ncbi.nlm.nih.gov

NCBI External Developer Release: SRA Toolkit 2.7.0
July 12, 2016

  align, bam-load: Insert-only alignments no longer cause incorrect binning
  bam-load: fixed case where WGS accessions where not being read correctly
  bam-load: will NOT perform spot assembly using hard clipped secondary alignments, even when 'make-spots-with-secondary' is enabled; WILL perform spot assembly using hard-clipped secondary alignments when 'defer-secondary' is enabled
  blast, kfg, ncbi-vdb, sra-tools, vfs: restored possibility to disable local caching
  build, sra-tools: Running make and make install in sra-tools repository prepares all configuration files required to access NCBI repository
  doc, ncbi-vdb: created a Wiki page illustrating how to use the API to set up logging
  fastdump, sra-tools: new tool to perform fast dumps of a whole run in either fastq or a custom format for one of our customers.
  kar: Alter the default ordering of components of an SRA archive for better network performance
  kdb, kfs, kns: Added blob validation for data being fetched remotely
  kfg: When loading configuration files on Windows USERPROFILE environment variable is used before HOME
  kfg: modified auxiliary remote access configuration to use load-balanced servers
  kns: Fixed a bug when KHttpRequestPOST generated an incorrect Content-Length after retry
  ngs, search, sra-search: sra-search was modified to support multiple threads.
  ngs-engine, ngs-tools, sra-tools, vfs: The "auxiliary" nodes in configuration are now ignored
  pileup-stats: updated commandline parser to eat unprocessed parameters
  sam-dump: updated to append asterisks to quality field of SAM output when corrupt original BAM has secondary alignment is shorter than the primary.
  search: now supports multi-threaded search 
  sra-search: now supports sorted output
  sra-tools: added possibility to build rpm package in sra-toolkit
  sra-tools: fixed exit codes for a number of applications in response to command line options
  vdb-dump: added tests to verify vdb-dump operation on nested databases
  vdb-validate: A new checks were added for SEQUENCE table
  vdb: fixed a bug in VCursorFindNextRowIdDirect where it returned a wrong rowId
  vdb: fixed a bug in the code used to iterate over blobs where rowmap expansion cache would reset iteration to initial row instead of respecting sequence
  vfs: environment variable VDB_PWFILE is no longer used


NCBI External Developer Release: SRA Toolkit 2.6.3
May 25, 2016

  bam-load: Corrects an optimization used to compare read lengths when lengths are greater than 255
  bam-load: alignments which are below the minimum match count but with at least 1/2 of the aligned bases matching are accepted
  bam-load: improved performance of SAM parsing code
  bam-load: non-fatal result codes no longer cause the reader thread to quit early
  bam-load: will NOT do spot assembly using hard clipped secondary alignments even when make-spots-with-secondary is enabled;
            WILL do spot assembly using hard-clipped secondary alignments when defer-secondary is enabled
  build: MSVS 2013 toolset (12.0) is now supported across all repositories
  vdb: Fixed a bound on memory cache that would never flush under certain access modes


NCBI External Developer Release: SRA Toolkit 2.6.2
April 20, 2016

  align-cache: a tool producing vdbcache that reduces effects of random access and improves speed.
  bam-load: Fixed bug in validation code for unsorted bam files
  bam-load: If two (or more) local reference names refer to the same global reference,
    bam-load will record the first one used and report the change.
  bam-load: Secondary alignment will be used for spot assembly if the loader is explicitly told to use them
  bam-load: The code that had invalid item in index problem was removed in the process of addressing a performance issue.
  bam-load: change reporting of fatal warnings into fatal errors
  bam-load: changed from an error to a warning if using secondary alignments to create spots
  bam-load: low-match secondary alignments are now discarded; low-match primary alignments are logged,
    and if too many, it will abort the load.
  bam-load: records the contents of the BX tag
  bam-load: rules for spot assembly were reverted to assembling spots only on primary alignments
  blast, build: Improved blast tools: all required libraries are linked statically. 
  build: Allow to build sra-tools on systems without static c++ library
  check-corrupt, vdb-validate: Added a new set of checks that can be triggered by using one of two "--sdc:" cmd options
  copycat: now runs on Centos 7; no longer uses system-installed magic file
  dbgap-mount: Added support for standard options  "-L" and "-o", which allow users
    to determine the logging level and log output file
  dbgap-mount: New optional parameter was introduced '-u' which allows user
    to unmount FUSE/DOKAN volume. Implemented on linux and windows
  fastq-load: The (old) fastq-loader will properly report multiple reads on the 454 platform (and still fail).
  kar: added '--md5' option to create md5sum compatible auxiliary file
  kdb: Fixed VTableDropColumn, so that it can drop static columns
  kfs, kns, ngs, sra-tools: Fixed thread safety issues for both cache and http files
  kget: kget has a new option --full to match wget in speed. added examples.sh
  kproc: Fixed KQueue to wake waiters when sealed, fixed KSemaphore to wake waiters when canceled
  latf-load: now allows undescores inside spot group names 
  latf-load: now loads data produced by fastq-dump
  latf-load: updated to support Illumina tag line format with identifier at the front
  pileup-stats: added -V (--version) option: prints out the software
  pileup-stats: pileup-stats: added version support (options -V or --version)
  prefetch: Added --eliminate-quals option which speeds up HTTP download by ignoring QUALITY column`s data
  prefetch: Fixed failure when running prefetch.exe "-a<bin|key>" when there is a space after "-a"
  prefetch: messages about maximum size of download are made more user-friendly.
  prefetch: now will download even when caching is disabled
  ref-variation: --input-file option allows to specify input accessions and paths in the file
  ref-variation: added "count-strand" option: it controls relative orientation of 3' and 5' fragments.
  ref-variation: added -c option to flush output immediately; reporting zero matches
  ref-variation: added a way to specify a number of repeats of the query
  ref-variation: improved threading management
  ref-variation: removed irrelevant warnings reported in some cases in debug version only
  sam-dump: Segfault no longer occurs when confronted with large amounts of header lines
  sam-dump: added option to produce MD tags
  sam-dump: filters out duplicates in the rows that it generates
  sam-dump: produces BX-tags if preserved in SRA file by bam-load
  sra-sort: correctly generates spot-id column even in the absence of primary alignments
  sra-stat: no longer fails when CS_NATIVE column is not present.
  sra-tools, vdb-config: Removed dependency of mac binaries on unnecessary libraries, e.g. libxml2.
  sra-tools: https://github.com/ncbi/sra-tools/issues/27 : contains short and long examples of how to configure sra-tools build
  var-expand: a tool for batch variation expansion
  vdb-config: now handles standard options such as --option-file
  vdb-validate: Added code to continue with the next row when column has discontiguous blobs

