SCJ-CARP is a prototype solving CARP, an ancestral reconstruction problem for pangenomes, under the SCJ-model. SCJ-CARP allows to calculate the SCJ-CARP measure, an index of pangenome structural complexity.
A necessary preprocessing step for the current version is some form of collinear block detection (for example SibeliaZ) and translating the block orders for each genome into unimog format.
Note: scj_carp depends on networkx.
Run python3 scj_carp.py <unimog> where <unimog> contains all genomes in unimog format (https://bibiserv.cebitec.uni-bielefeld.de/dcj).
The script will output a triple ("Carp index") consisting of the CARP measure, total number of marker occurrences in the pangenome and relative CARP measure (CARP measure divided by pangenome size).
| parameter | purpose |
|---|---|
--write-measure <file> |
Write the SCJ-CARP index to <file> |
--write-carp-adjacencies <file2> |
Write one possible set of adjacencies of an SCJ-CARP ancestor to <file2> |
--core |
Project the set of markers to only those occurring in all extant genomes. (Currently not recommended) |
The workflow depends on both Snakemake and ZOMBI. Be sure that both of these are installed on your system.
Important: Before running the workflow change the variable ZOMBI_HOME in the Snakefile to the installation of ZOMBI on your machine.
To replicate the experiments from the manuscript, first navigate in your command line to the top directory of this repository. You can then run the two experiments with the following rules:
| Experiment | Rule | Result File | Plot Scripts (./plotscripts/) |
|---|---|---|---|
| Scaling rates of Adjacency modifying operations (Duplications, Losses, Inversions, Transpositions, Originations) | all_treescale_results |
treescale/treescale.txt |
rearrangement_vs_measure.py, rearrangement_vs_precrec.py |
| Scaling rates of Non-Adjacency modifying operations (Horizontal (Replacement) Transfers) | all_transfer |
transferexp/transfercarp.txt |
transfer_vs_measure.py, transfer_vs_precrec.py |
To run the workflow, simply type snakemake <rule> --cores <n-cores>, where <rule> is one of the above rules and <n-cores> is the number of cores you want to use.
For example, snakemake all_treescale_results --cores 4.
The table of results (SCJ-CARP measure) and precision/recall can then be found in the corresponding result file. To replicate one of the pots, simply run the corresponding script, inputting the result file as an argument, for example
python3 plotscripts/rearrangement_vs_precrec.py treescale/treescale.txt.