Skip to content

Genome assembly scaffolding using information from paired-end/mate-pair libraries, long reads, and synteny to closely related species.

License

Notifications You must be signed in to change notification settings

lpryszcz/pyScaf

Repository files navigation

pyScaf orders contigs from genome assemblies utilising several types of information:

This is under development... Stay tuned.

In this mode, pyScaf aligns long reads onto the contigs, identifies the reads the connects two or more contigs and join adjacent contigs.

Long reads are aligned locally onto contigs, ignoring:

  • matches not satisfying cut-offs (--identity and --overlap)
  • suboptimal matches (only best match of each query to reference is kept)
  • and removing overlapping matches on reference.

Note, this is experimental implementation.

In reference-based mode, pyScaf uses synteny to the genome of closely related species in order to order contigs and estimate distances between adjacent contigs.

Contigs are aligned locally onto reference chromosomes, ignoring:

  • matches not satisfying cut-offs (--identity and --overlap)
  • suboptimal matches (only best match of each query to reference is kept)
  • and removing overlapping matches on reference.

In preliminary tests, pyScaf performed superbly on simulated heterozygous genomes based on C. parapsilosis (13 Mb; CANPA) and A. thaliana (119 Mb; ARATH) chromosomes, reconstructing correctly all chromosomes always for CANPA and nearly always for ARATH (Figures in dropbox, CANPA table, ARATH table). Runs took ~0.5 min for CANPA on 4 CPUs and ~2 min for ARATH on 16 CPUs.

Important remarks:

  • Reduce your assembly before (fasta2homozygous.py) as any redundancy will likely break the synteny.
  • pyScaf works better with contigs than scaffolds, as scaffolds are often affected by mis-assemblies (no de novo assembler / scaffolder is perfect...), which breaks synteny.
  • pyScaf works very well if divergence between reference genome and assembled contigs is below 20% at nucleotide level.
  • pyScaf deals with large rearrangements ie. deletions, insertion, inversions, translocations. Note however, this is experimental implementation!
  • Consider closing gaps after scaffolding.

Given reference genome, the program generates pairwise genome alignment (dotplots) by default.

  • Genral options:

    -h, --help

    show this help message and exit

    -f FASTA, --fasta FASTA
     

    assembly FASTA file

    -o OUTPUT, --output OUTPUT
     

    output stream [scaffolds.fa]

    -t THREADS, --threads THREADS
     

    max no. of threads to run [4]

    --log LOG

    output log to [stderr]

    --dotplot

    generate dotplot as [png]

    --version

    show program's version number and exit

  • Reference-based scaffolding options:

    -r REF, --ref REF, --reference REF
     

    reference FastA file

    --identity IDENTITY
     

    min. identity [0.33]

    --overlap OVERLAP
     

    min. overlap [0.66]

    -g MAXGAP, --maxgap MAXGAP
     

    max. distance between adjacent contigs [0.01 * assembly_size]

    --norearrangements
     

    high identity mode (rearrangements not allowed)

  • Long read-based scaffolding options (EXPERIMENTAL!):

    -n LONGREADS, --longreads LONGREADS
     

    FastQ/FastA file(s) with PacBio/ONT reads

  • NGS-based scaffolding options (!NOT IMPLEMENTED!):

    -i FASTQ, --fastq FASTQ
     

    FASTQ PE/MP files

    -j JOINS, --joins JOINS
     

    min pairs to join contigs [5]

    -a LINKRATIO, --linkratio LINKRATIO
     

    max link ratio between two best contig pairs [0.7]

    -l LOAD, --load LOAD
     

    align subset of reads [0.2]

    -q MAPQ, --mapq MAPQ
     

    min mapping quality [10]

To perform reference-based assembly, provide assembled contigs and reference genome in FastA format. Dotplots of below runs can be found in docs. If you wish to skip dotplot generation (ie. no X11 on your system), provide --dotplot '' parameter.

# scaffold homogenised assembly (reduced contigs)
./pyScaf.py -f test/contigs.reduced.fa -r test/ref.fa -o test/contigs.reduced.ref.fa

# scaffold reduced contigs using global mode (no norearrangements allowed)
./pyScaf.py -f test/contigs.reduced.fa -r test/ref.fa -o test/contigs.reduced.ref.global.fa --norearrangements

# scaffold heterozygous assembly (de novo assembled contigs)
./pyScaf.py -f test/contigs.fa -r test/ref.fa -o test/contigs.ref.fa

# scaffold reduced contigs using long reads
## pacbio
./pyScaf.py -f test/contigs.reduced.fa -n test/pacbio.fq.gz -o test/contigs.reduced.pacbio.fa
## nanopore
./pyScaf.py -f test/contigs.reduced.fa -n test/nanopore.fa.gz -o test/contigs.reduced.nanopore.fa

# generate dotplot
lastdb test/ref.fa
lastal -f TAB test/ref.fa test/contigs.reduced.pacbio.fa | last-dotplot - test/contigs.reduced.pacbio.fa.ref.png
lastal -f TAB test/ref.fa test/contigs.reduced.nanopore.fa | last-dotplot - test/contigs.reduced.nanopore.fa.ref.png

# clean-up
#rm test/contigs.{,reduced.}fa.* test/ref.fa.* test/*.{nanopore,pacbio,ref}* test/*.log

pyScaf is under heavy development right now. Nevertheless, both the reference-based mode and long-read mode are functional and produces meaningful assemblies. pyScaf has been implemented in Redundans.

For more info, have a look in workbook.

About

Genome assembly scaffolding using information from paired-end/mate-pair libraries, long reads, and synteny to closely related species.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages