odgi

List of commands is available here.

Warning

To install the latest version of odgi (0.8.6 at the time of writing) through conda, make sure to setup an environement with the latest version of python (>=3.11) and libgcc-ng >=12, otherwise an older version of odgi will be installed!

conda create -n odgi python=3.11
conda activate odgi
conda install -c conda-forge libgcc-ng
conda install -c bioconda odgi

General commands §

Convert graph formats §

To convert from odgi (.og) format to another formats (like GFA for instance) it is possible to use odgi view.

⚠️ If you want to have the paths in your .og file (and any conversion downwards), you need to give a GFA1.0 file to odgi (and not a GFA1.1 or rGFA).

# Convert to GFA
odgi view -g -i $INPUT > $OUTPUT
# INPUT is a .og file
# OUTPUT is a new .gfa file
# -g stands for "convert to gfa"

To convert from GFA to odgi, you can use odgi build

Draw a static render of a graph §

To create a static (.png) representation of a graph, two steps are required. The first one is to layout the graph with odgi layout to get a layout file (where to position nodes in the static file) than using odgi draw to render the final picture. First and foremost, before the layout computation, the graph has to be optimized. Interestingly, on graphs from 42 T2T yeast genomes, the graph from minigraph-cactus was optimized but not the one from pggb.

# You can convert the graph in .og format to speed up the process
odgi build -g graph.gfa -o graph.og
# If needed, optimize the graph
odgi sort -i graph.og -O -o graph_opti.og
# Create the layout
odgi layout -i graph.og -o layout.lay
# Draw the final figure
 

Extract subgraph §

It exists many criterion to extract a graph in odgi. The command odgi extract allows for queries based upon positions, nodes and ranges. To extract an genomic interval, one can use:

odgi extract -i input.gfa -o output.og --pangenomic-range XXX-YYY

To extract an interval along a path, one can use:

odgi extract -i input.gfa -o output.og --path-range PATHNAME:XXX-YYY

Coordinates are give as zero-based coordinates (starting at 0 at the start of the path)

To extract a list of nodes: create a .txt file and write nodes names, one per line. Then,

odgi extract -i input.gfa -o output.og --node-list file.txt

Useful supplementary arguments:

--context-steps n # Extracts n surrounding nodes from the query
--context-bases n # Extracts n surrounding bases from the query
--paths-to-extract file.txt # Keeps only the given paths

Break cycles in the graph §

Cycles can harm capacitiy to run certain algorithms on pangenome graphs. Odgi has a tool, odgi break that destroys loops. However it drops paths!

Python bindings §

Warning

It exists an older implementation of bindings, which is the one referenced in the readsthedocs.io, HOWEVER it is not the one which should be used for performance reasons as well as stability issues…

According to the documentation, odgi_ffi is meant to be used more as a tool to build a Python library than being the actual Python library.

Note that odgi also has an older high-level Python API import odgi that is somewhat obsolete. Instead you should probably use below import odgi_ffi lower level API to construct your own library.

In order to fix segfaults, set LD_PRELOAD=libjemalloc.so.2 before running Python scripts. However, I could not get it to work in any way, as if I can as the time I’m writing those lines (dec. 2023) import the bindings, I could not load a graph, giving an error when I try to (RuntimeError: Error rewinding to load non-magic-prefixed SerializableHandleGraph) . Given the maintainers are not implying that the library is not fully stable I won’t settle on it for the future.