Magic-BLAST 1.6.0: May 6, 2021
- Usage reporting - Help improve Magic-BLAST by sharing limited information about your search. Details on the information collected, how it is used, and how to opt-out.
- Reads from NCBI SRA repository are streamed from the cloud when
-sra_batchoption is used. More details here.
- NCBI taxonomy IDs are reported in SAM output if they are present in the target BLAST database.
- Unaligned reads can be reported separately from the aligned ones, using
-out_unaligned <file name>option. One can also select the format with
-unaligned_fmtoption. Choices are SAM, tabular, and FASTA. The default format is the same as one for the main report (SAM or tabular, selected with
- A file with list of SRA accessions can be provided to Magic-BLAST via the
- Magic-BLAST correctly reports database sequence accessions for BLAST databases without gis.
Fixed discontinuity in adaptive score threshold function. Below are the new thresholds:
Read length (r) Score threshold r <= 20 r 20 < r <= 34 20 50 < r < 200 0.6 * r r >= 200 120
Magic-BLAST 1.5.0: August 22, 2019
- Support for the new BLAST database version (BLASTDBv5) that allows for limiting search by taxonomy (more information about database version 5 here)
- New option
-md_tag: SAM MD tag is no longer reported by default. To have it included in SAM report, use
- New symbol in tabular report BTOP string:
%<number>%that represents a deletion (gap in read) of this number of bases.
New adaptive alignment score threshold, calculated based on read length (score thresholds below). This is the default behavior. Users can change alignment score threshold with the
-scoreoption and set it either to a constant or a linear function of read length.
Read length (r) Score threshold r <= 20 r 20 < r <= 30 20 30 < r <= 50 r - 10 50 < r < 200 0.6 * r r >= 200 120
- Improved multi-threading and run time.
- Improved alignment heuristics that allow for larger error rates and better alignments for long reads.
- Magic-BLAST aligns nanopore reads.
- NCBI accessions instead of gis are reported in SAM and tabular reports.
- Short, low-complexity alignments are no longer reported.
- The default value for
-max_db_word_countparameter was lowered to from 60 to 30. 16-base words that appear in the genome more than this number of times will be filtered.
- The maximum insert size for properly aligned pairs is 1,000,000 bases for spliced alignments (RNA-seq) and 100,000 bases for non-spliced alignments (genomic). The alignments for pairs with larger insert size are still reported, but SAM flag for properly aligned pair is not set.
- SAM MD tag reports correct number of matching bases around an intron.
-max_db_word_countoption no longer requires explicit use of
- Magic-BLAST no longer crashes with an empty sequence in FASTQ file.
Magic-BLAST 1.4.0: August 10, 2018
- New option:
-no_discordantto report only concordant read pairs
- Report strand-specific alignments with
-rfflags for forward-reveresed and reversed-forward
- New option to control repeat filtering:
-max_db_word_count. 16-base words that appear in the genome more than this number of times will be filtered (default is 60).
- Improved sensitivity: count for frequent database words to be filtered was increased to 60 (used to be 10). This threshold can be changed with the
- Non-cannonical splice signal now require longer alignments on the exon, with score at least 50 on both sides of an intron.
- More informative error messages for SRA access
- Much better handling of non-cannonical splice sites and compositionally biased genomes
- Alignments no longer stop prematurely
- Fix for not returning unmapped reads when none is aligned
- Magic-BLAST no longer reports zero-length introns
- Parameters of the score threshold as a function of read length are no longer 100 smaller than specified by the user
Magic-BLAST 1.3.0: September 15, 2017
- The alignment cutoff score (
-scoreoption) can be expressed as either a constant or a function of read length in this format: L,b,a for a * length + b
- Maximum edit distance cutoff for alignments can be specified with
- SRA caching is now turned off by default and can be turned on with the
-sra_cacheflag. When accessing sequences in NCBI SRA repository the data can be cached in a local file so that it is not downloaded over the network again when reused later. It may result in very large local files and is only needed if you use SRA accessions multiple times, have very limited network bandwidth and a large disk space.
- Unmapped reads are reported in the SAM and Tabular reports, use
-no_unalignedoption to not have unaligned reads reported
- The XS tag (used by transcript assemblers) is now reported in SAM output
- The score threshold (
-scoreoption) now applies to the whole spliced alignment
- The query batch size (number of reads processed at a time) is now controlled with environment variable BATCH_SIZE expressed in cumulative number of bases
- The default mismatch and gap extension penalties are now set to 4
- Improved sensitivity and run time
- Read ids for pairs are printed properly in the SAM format (no “.1” “.2” or “/1” “/2”)
- The secondary alignment bit (256) is set in SAM flags
- Maximum intron length option
Magic-BLAST 1.2.0: February 17, 2017
- Improved multi-threading for larger genomes
- Improved splice site detection
- Magic-BLAST now works with multiple SRA accessions
- Fixed the macOS dmg installer that used to remove BLAST+ binaries
- The -seqidlist option is no longer ignored
Magic-BLAST 1.1.0: November 4, 2016
- -sra option connects to NCBI via HTTPS
- Results are formatted with ‘bare’ accessions
- Tabular output includes a header with column titles
- Fixed SAM flag values
Magic-BLAST 1.0.0: August 19, 2016
- First release