converts (sorted) genome variation graphs (in GFA or .xg format) to multiple alignment format (MAF)
This tool projects between pangenomic variation graphs stored in GFAv1, which can be used to encode whole genome alignments, and the multiple alignment format MAF, which represents only the linearizable components of such an alignment graph.
Its goal is to allow tools like seqwish, which efficiently construct whole genome alignment graphs from collections of sequences and pairwise alignments between them, to be applied to comparative genomics problems.
Graphs in GFA must be encoded as is standard in variation graph based methods.
We use the S, L, and P records.
S encodes sequences, whose identifiers should be numeric values.
L encodes links between sequences.
P describes the paths of genomes through the graph.
Graphs of this format are produced by seqwish.
The graph should be sorted in a reasonable way, which can be accomplished with odgi sort.
This is because the blocks in the emitted MAF format are linear components in the sorted graph.
We determine blocks by finding regions in which all path positions versus the sorted graph space are monotonically increasing or decreasing.
In short, nonlinear structural variation (like collapsed CNVs) and rearrangements will break the output blocks, as these cannot be represented with a gapped multiple alignment.
maffer uses cmake to build itself and its dependencies.
git clone --recursive https://github.com/pangenome/maffer.git
cd maffer
cmake -H. -Bbuild && cmake --build build -- -j 4
Conversion to MAF can be obtained in a single step:
maffer -g graph.gfa -m >aln.maf
Alternatively, an .xg index can be used. This matters for larger graphs, where it can make it quicker to iterate different MAF projection parameters.
maffer -o graph.xg -g graph.gfa
maffer -i graph.xg > aln.maf
MIT
Erik Garrison
Simon Heumos