Rules

The pipeline contains the following endpoint rules:

ggcaller
pangenome
annotate_pangenome
lineage_st
find_amr_vag
tree
unitigs
panfeed
combine_heritability
pyseer
pyseer_rare
wg
wg_metrics
map_back
qq_plots
manhattan_plots
enrichment
enrichment_plots
annotate_summary

Which accomplish the following functions:

  • ggcaller: will generate GFF annotations for all sample genomes using ggCaller, providing gene predictions and functional annotations. If pre-computed GFF files are specified via the input table, this rule will instead validate the user-provided paths and skip the time-consuming de novo gene calling step.

  • unitigs: will generate a variant set from the input samples based on a “global” de Brujin graph.

  • lineage_st: will generate a tab-separated file with the predicted sequence types (STs).

  • find_amr_vag: will generate a summary file with the predicted antimicrobial resistance genes identified into functionally relevant groups, and a summary file with the predicted virulence associated genes.

  • pangenome: will find the orthologous gene clusters across all samples and the chosen references using panaroo.

  • annotate_pangenome: will generate functional annotations for the pangenome using eggnog-mapper, including COG categories, GO terms, and KEGG pathway mappings.

  • tree: will generate a phylogenetic tree from the core genome alignment output from panaroo.

  • combine_heritability: will generate a file with the comined heritabilities: built from the lineages of each strain and by using a kinship matrix built from the unitigs presence and absence matrix.

  • pyseer: will test for associations of each unitig and the phenotype, as well as gene presence/absence patterns and lineage effects.

  • pyseer_rare: will test for rare variants based on the predicted deleterious protein coding variants.

  • panfeed: will test gene-cluster specific k-mers for their association with the phenotype(s), and produce output plots.

  • wg: will train two machine learning models (lasso and a ridge elastic nets) based on the presence/absence patterns of all unitigs.

  • wg_metrics: will calculate prediction performance metrics for the machine learning models trained by the wg rule, providing quantitative assessment of model accuracy.

  • qq_plots: will createa Q-Q plot to check that p-values are not inflated (large ‘shelves’ are symptomatic of poorly controlled confounding population structure)

  • map_back: will map back the associated genetic variants to the provided reference genomes.

  • manhattan_plots: will generate a Manhattan plot of the unitigs that map to the chosen reference genome.

  • annotate_summary: will generate an annotated summary table for all associations, including: the identity of the gene the variants map to, the number of strains, the average association pvalue, the gene ID across the selected reference genomes, and automatic annotations provided by eggnog-mapper.

  • enrichment: will generate a file with the functional enrichment of the associated variants for GO terms, COG categories and KEGG pathways.

  • enrichment_plots: will generate visualizations from the results of the enrichment rule.

Tip

Please note that some of the above rules will depend on each other.