Building an index

Given a genome sequence file "genome.fasta", you can build an index by issuing the following command:

$ ./hobbes-index --sref genome.fasta -i genome.hix -g 11 -p 4

The command makes an index file named "genome.hix" using gram length 11. hobbes-index uses --sref to specify an input genome sequence file, -i to specify output index file, and -g to specify the gram length. -p enables multithreading and the command builds an index using 4 threads.

* Hobbes2 builds an index very fast. For the whole human genome of HG18, for example,
  the command above built an index in 9 minutes on a machine with 94 GB of RAM, and
  dual quad-core Intel Xeons X5670 at 2.93 GHz, running a 64-bit Ubuntu OS.

Aligning single-end reads

Given a read file "read.fastq" and a genome sequence file "genome.fasta" with its index file "genome.hix", the following command finds all mappings within edit distance 5.

$ ./hobbes -q read.fastq --sref genome.fasta \
           -i genome.hix -a --indel -v 5 -p 16 --mapout output.sam

hobbes uses -q to specify an input read file, --sref to specify a genome file, and -i to specify an index file for the genome file. By using -a, you can make hobbes produce all mapping locations.--indel indicates gapped alignment with an edit distance threshold of 5, which is specified by -v. -p enables multithreading and the command aligns reads using 16 threads. hobbes produces results in the SAM format. The command output results to a file named "output.sam", which is specified by --mapout. If --mapout is not specified, hobbes outputs results to stdout. In this case, you can redirect results to an output file as follows.

  
$ ./hobbes -q read.fastq --sref genome.fasta \
           -i genome.hix -a --indel -v 5 -p 16 > output.sam

If you want to align reads with a Hamming distance threshold instead of an edit disance, you can replace --indel with --hamming as follows.

$ ./hobbes -q read.fastq --sref genome.fasta \
           -i genome.hix -a --hamming -v 5 -p 16 --mapout output.sam

If you specify the number of reads, N, using -n, hobbes maps only the first N reads from the input read file. By using -n, you can see the progress with an estimated time to complete alignment. It can be useful for testing your pipeline.

$ ./hobbes -q read.fastq --sref genome.fasta \
           -i genome.hix -a --indel -v 5 -n 10000 -p 16 --mapout output.sam

76% MAPPING READS: 7616/10000; 0'26"/0'35"

Aligning paired-end reads

To align paired-end reads, you should use --pe and specify two input read files using --seqfq1 and --seqfq2, respectively. You also need to specify minimum insert size and maximum insert size using --min and --max, respectively. Other options are exactly the same as those of single-end alignment. Given two read files "read1.fastq" and "read2.fastq", for example, the following command produces paired-end mappings.

$ ./hobbes --pe --seqfq1 read1.fastq --seqfq2 read2.fastq --min 110 --max 290 \
           --sref genome.fasta -i genome.hix -a --indel -v 5  -n 10000 -p 16  \
           --mapout output.sam