audy / stitch

Overlap assembler of paired-end DNA sequences generated by Illumina
MIT License
22 stars 7 forks source link

Stitch

No longer maintained. Please use Pandsaseq.

Austin G. Davis-Richardson harekrishna@gmail.com

Stitch assembles overlapping paired-end reads into a single contig for each pair. This increases the read length and hopefully the quality of a de novo or reference assembly. Stitch is multi-threaded and will automatically use all cores on your system unless told otherwise. Stitch currently only reads FASTQ format. QSEQ and FASTA formats to come. Reads that are not found to overlap are dumped in a file called <prefix>-singletons and are in FASTQ format. These can then be trimmed and combined with contigs to do a de novo assembly.

Stitch is licensed under the MIT open source license. See LICENSE.md for details.

Citing Stitch

Please us the following citation:

Brown, Christopher T., et al. "Gut microbiome metagenomics analysis suggests a functional model for the development of autoimmunity for type 1 diabetes." PloS one 6.10 (2011): e25792.

Or, in BibTex:

@article{brown2011gut,
  title={Gut microbiome metagenomics analysis suggests a functional model for the development of autoimmunity for type 1 diabetes},
  author={Brown, Christopher T and Davis-Richardson, Austin G and Giongo, Adriana and Gano, Kelsey A and Crabb, David B and Mukherjee, Nabanita and Casella, George and Drew, Jennifer C and Ilonen, Jorma and Knip, Mikael and others},
  journal={PloS one},
  volume={6},
  number={10},
  pages={e25792},
  year={2011},
  publisher={Public Library of Science}
}

Requirements

Alignment Algorithm

Stitch aligns overlapping paired end reads by counting the number of matching nucleotides in an overlapping window. The window that provides the highest number of matching nucleotides wins.

The consensus sequence is generated thusly,

A 5' =============== =================================> 3'
B                3' <================================= ===========  5'
C 5' =============== ================================= ===========> 3'
                    <--------- "the middle" ---------->

In the region dubbed "the middle", the nucleotide with the highest corresponding quality score is used (if there is a mismatch). If there is a match, then that nucleotide is used (and the highest quality score is given). In the case of a tie, a 'N' is used and the quality score is unchanged.

Stitch assumes that read A is 3'-5' and read B is 5'-3'. So read B is automatically reverse-complemented in the alignment procedure. (I know I should make this an option but I haven't yet).

Usage

NOTE - Stitch expects reads to be of the same length!

Invoke, comme ca

Usage: stitch.py -i <fastq file 1> -j <fastq file 2> -o <output prefix>

More options,

-h, --help            show this help message and exit
-i FILEA, --first=FILEA
                    first fastq file
-j FILEB, --second=FILEB
                    second fastq file
-o PREFIX, --output=PREFIX
                    output prefix (omit to print to stdout)
-t THREADS, --threads=THREADS
                    number of threads (default = all available)
-p, --pretty_output   displays overlapping contigs in a nice way.
-s SCORE, --score=SCORE
                    minimum percent identity (default = 25)
-b TABLE, --table=TABLE
                    output overlap length to a text file

Bugs/Feature requests