mafviewer

A JBrowse 1 plugin for viewing multiple alignments.

Note: See https://github.com/cmdcolin/jbrowse-plugin-mafviewer for JBrowse 2 version!

Prepare data

This program supports two formats

1. BigMaf format, which can be created following UCSC guidelines

2. MAF tabix based format, based on a custom BED created via conversion tools in this repo.

The choice between the two is your convenience. BigMaf is a "standard" UCSC format, basically just a specialized BigBed, so it requires JBrowse 1.14.0 or newer for it's BigBed support. The custom BED format only requires JBrowse 1.12.3 or newer, so therefore some slightly older JBrowse versions can support it.

Note: Both formats start with a MAF as input, and note that your MAF file should contain the species name and chromosome name e.g. hg38.chr1 in the sequence identifiers.

Preparing BigMaf

Follow instructions from https://genome.ucsc.edu/FAQ/FAQformat.html#format9.3 and set the storeType of your track as MAFViewer/Store/SeqFeature/BigMaf

Preparing the tabix BED format

Start by converting the MAF into a pseudo-BED format by calling bin/maf2bed.pl

bin/maf2bed.pl hg38 < file.maf > output.txt
bgzip output.txt
tabix -p bed output.txt.gz

Note: currently assumes space separated MAF, not tab separated. If you have tab separated, can add a 'sed' command in pipe

cat file.maf | sed -e 's/\t/ /g | ./maf2bed.pl hg38 | bgzip > out.txt.gz

The second argument to maf2bed.pl is the genome version e.g. hg38 used for the main species in the MAF (if your MAF comes from a pipeline like Ensembl or UCSC, the identifiers in the MAF file will say something like hg38.chr1, therefore, the argument to maf2bed.pl should just be hg38 to remove hg38 part of the identifier. if your MAF file does not include the species name as part of the identifier, you should add the species into them the those scaffold/chromosome e.g. create hg38.chr1 if it was just chr1 before)

If all is well, your BED file should have 6 columns, with chr, start, end, id, score, alignment_data, where alignment_data is separated between each species by ; and each field in the alignment is separated by :.

Note: you can also stream from a gzipped MAF to the bgzipped bed

gunzip -c chr21.maf.gz | bin/maf2bed.pl hg38 | bgzip > output.txt.gz

The bin/convert.sh script has a small automatic processing from maf to bgzipped, tabixed, bed.

Options

• samples - an array of species in the MAF file (e.g. hg38, mm10, etc.)
• labelWidth - an integer width for labels (default: 100)
• style.matchColor - color to use for matches (default: green)
• style.mismatchColor - color to use for mismatches (default: blue)
• style.gapColor - color to use for gaps in alignment (default: red)
• style.mismatchBases - set to true, then you can set style.mismatchA, style.mismatchG, style.mismatchC, style.mismatchT as needed
• storeClass - set to MAFViewer/Store/SeqFeature/MAFTabix or MAFViewer/Store/SeqFeature/BigMaf

Note: samples can be the array of strings like ["hg38","mm10"] or an array of extended JSON structures in order to customize the subtrack labels e.g. {"id": "hg38", "label": "Human", "description": "Extended description of species", "color": "rgb(255,255,0)" }

Example config

{
  "label": "MAF",
  "urlTemplate": "chrI.txt.gz",
  "storeClass": "MAFViewer/Store/SeqFeature/MAFTabix",
  "type": "MAFViewer/View/Track/MAF",
  "samples": [
    "cb4",
    "caeRem4",
    "caePb3",
    "caeSp111",
    "caeJap4"
  ]
}

Screenshot

For comparison this is the same region in UCSC browser (picture) (link)

Installation

Download to the plugins/MAFViewer and add to your config file with

"plugins": ["MAFViewer"]

See JBrowse FAQ on installing plugins

Demo

The test/ directory contains sample data for C. Elegans (from UCSC), Human (from UCSC), Medaka (from Ensembl).

Visit http://localhost/jbrowse/?data=plugins/MAFViewer/test/data or http://localhost/jbrowse/?data=plugins/MAFViewer/test/medaka or http://localhost/jbrowse/?data=plugins/MAFViewer/test/hg38 to view

Pairwise alignment

You can obtain MAF output from lastz using --format=maf and then use mafviewer to compare two different genomes

You can also obtain MAF from mummer by converting the outputted .delta file to MAF with the delta2maf program (not distributed in the latest mummer versions, you can obtain delta2maf by downloading mugsy from sourceforge and finding delta2maf inside their version of mummer. Then run delta2maf yourfile.delta > yourfile.maf note that delta2maf assumes the fasta files are located where the .delta line 1 says they are)

Important: if using MAF files from lastz or mummer here, you should edit the MAF to include the organisms name pre-pended onto the chromosome names, e.g. if it says chr1, add "human.chr1" where relevant. Then the bin/maf2bed.pl program included in this package can be run with bin/maf2bed.pl human < yourfile.maf > output.bed which then strips the "human" part of the chromosome identifiers again

Multiple alignment

For performing multiple alignment, https://github.com/medvedevgroup/SibeliaZ is an option

Alternatively, you can use the classic TBA/Multiz pipeline from UCSC, e.g. http://shiulab.plantbiology.msu.edu/index.php?title=SyntenicBlocks. I have not evaluated either of these pipelines, but the main goal is that resulting alignments must be non-overlapping for mafviewer to properly display them

Notes

Requires JBrowse 1.12.3 or later for BEDTabix functionality

Requires the alignment chunks to be non-overlapping (it will plot overlapping chunks of the MAF file over each other on the same line see #7)

Feel free to provide feedback!