brat-1.2.4
In version 1.2.4 we fixed a bug in convert-to-sam. Bug did not affect correctness of the output. In previous version it was assumed that all options will be provided with a user (-p, -1, -2, -s), and in this version we fixed this.

brat-1.2.3
In version 1.2.3 we added a new tool convert-to-sam that converts BRAT format to SAM format.

brat-1.2.2
In version 1.2.2 we changed trim program slightly to allow for different length of mates in paired-end reads. In previous versions, trim expected the lengths of two mates of a paired-end read to be equal.

brat-1.2.1
In version 1.2.1 we changed input reads format as well as output format to allow BRAT to remove copy-duplicates. We added a new tool remove-dupl that removes copy-duplicates. Finally, we made changes to User Manual and changed the names of files with unmapped and ambiguous reads.

brat-1.1.20
In version 1.1.20, we slightly changed the output format for acgt-count with option -B; now the format contains content information for each cytocine (CG, CHG or CHH). We made some changes to section How to use BRAT with reads sequenced from four strands in USER MANUAL.

brat-1.1.19
In version 1.1.19, we fixed a bug that caused segmentation fault on some reads; we fixed a bug in acgt-count that caused incorrect output when using new option -B with this program; we fixed a bug in acgt-count that could cause incorrect ACGT-count when reads come from four strands of PCR-product (the bug did not affect the results if reads come from the original genomic strands).

brat-1.1.18
In version 1.1.18, we fixed a bug in brat-large: this bug caused segment fault when using brat-large with previously built index using brat-large-build. This bug did not affect mapping results when using brat-large with the list of references.
Please note that we made changes to USER MANUAL on How to Use BRAT with reads sequenced from four strands to make sure that acgt-count works correctly on four strands as well.

brat-1.1.17
In version 1.1.17, we fixed a bug in acgt-count: in the previous versions it had only option -s to accept the file with names of the files with mapping results of single-end reads; however, for paired-end reads, if users choose to map files with mates 2 as single-end reads, ACGT-distribution has to be counted differently than for mates 1 (please refer to section 5 for details). Hence, we added two new options to use with acgt-count (please see how to use ACGT-count) to distinguish between the two types of single-end reads (mates1 and mates 2 of paired-end reads). In addition, we added an additional option -B to use with acgt-count: with this option users will have only two output files in the different format than before (please refer to the section Output format for ACGT-count). Finally, we changed the output of check-strands. Now, instead of ACGT-counts it outputs the ratios: A/(A+C+G+T), C/(A+C+G+T), G/(A+C+G+T) and T/(A+C+G+T).

brat-1.1.16
In version 1.1.16, we added additional capability to BRAT: now it will map reads sequenced from four PCR-product strands. Please read details in Section 4, Details on ACGT-count, of User Manual.

brat-1.1.15
In version 1.1.15, we fixed a bug in acgt-count and included two new tools: rev-compl and check-strands. Recently, we received e-mails from the users whose questions made us re-think the way we count ACGT-distribution of mapped reads for positive and negative strands. Please read a full account of these changes in section 4 "Details on ACGT-count" of the User Manual. Some users believe that the sequenced reads are produced by sequencing all four strands (two bisulfite-treated original genomic strands and their reverse-complements resulted from PCR for library amplification). In section 4, we describe how to use BRAT to check whether the reads are sequenced from two original strands as templates or from four PCR-product strands as templates for sequencing. Finally, we provide instructions on using BRAT in the case when the reads are sequenced from four strands as templates.

brat-1.1.14
In version 1.1.14, we fixed a bug in tool acgt-count. If minimum insert size (option i) was set to be smaller than the minimum read length (in the set of input reads), then for paired-end mapping two mates could overlap each other. In previous versions, (A,C,G,T) count in the overlapping regions included nucleotide coverage from both mates. In this new version, nucleotide coverage in the overlapping region is counted only from one mate of a pair.

brat-1.1.13
In version 1.1.13, we fixed bugs in tools acgt-count, brat-large-build and brat-large . The bugs did not affect the correctness of mapping.

brat-1.1.12
In version 1.1.12, we added a new option, L, to use with tool trim to allow different base quality score types such as Phred or Solexa/Illumina.

brat-1.1.11
Fixed bug in trim: now it allows for a slightly different FASTQ input.
Added a new option, m, to tool trim to allow up to m internal Ns.
Now the tool trim has additional output files in FASTQ format to track original reads' names and corresponding base quality scores.

brat-1.1.10
In brat-1.1.10, we added minor changes to brat-large: a warning about the range for option f and a checking condition to ensure that f has a value between 24 and 64 inclusively.

brat-1.1.9 and brat-1.1.8
In brat-1.1.8, we corrected some conditions that caused the programs to stop trying to map a read. This affected only mapping with non-BS-mismatches. More reads are now mapped. The changes in brat-1.1.8 made brat unnecessarily slow, so we made additional changes in brat-1.1.9.

brat-1.1.7
In this version, we offer users two new options for use with BRAT-large: P and S. Option S gives the users control over (space usage)-(mapping time) balance, and the option P gives the users a choice of separating hashing of the genome from mapping reads to the genome. If users map reads to the same genome and wish to re-use the same hashing index, they now have the choice to pre-build the hashing index, save it on a hard disk and then re-use the index for mapping different reads. Mapping with P is done in two steps: first run the program brat-large-build and then run brat-large with option P.

brat-1.1.6
Further improved space usage for brat-large. It was [269M + (3/8)N + 4S] bytes and now it is [269M + (3/8)S + 4S] bytes, where N and S are sizes of a genome and the largest reference, respectively, in base pairs. On human genome with 10mln pairs, brat-large now uses less than 2GB.

brat-1.1.5

• Improved space-usage for both brat and brat-large. In the previous version of brat, space usage was [269M + (3/8)N + 2*8N] bytes, and now it is [269M + (3/8)N + 2*4N] bytes, where N is the genome's size in base pairs. For brat-large, space also decreased significantly: from [269M + (3/8)N + 8S] bytes to [269M + (3/8)N + 4S] bytes, where S is the size of the largest reference.
• A restriction on the number of references is removed and now the number of references is virtually unlimited.
• Some restrictions are removed from acgt-count to make this program more flexible.

brat-1.1.4

• Improved space-usage for brat: in brat-1.1.3, brat used 4*8N bytes (N is the genome's size in base pairs), and in brat-1.1.4, brat uses 2*8N bytes.
• A new option "f" is offered: for brat, it sets the number of the first bases of a read, where only one non-BS-mismatch is allowed; and for brat-large, it sets the number of the first bases of a read, where non-BS-mismatches are NOT allowed. This will allow users to speed up running time for longer reads (longer than 64 bases) if the users choose f greater than the default values (36 for brat and 24 for brat-large).