Modes

BacFlux has four front ends and one analysis. mode in the config picks the front end; nothing on the command line changes with it.

A front end is everything up to and including the assembly: read QC, the assembler, the contig filter and the contamination screen. It ends by writing one file, 02.assembly/{sample}/contigs_final.fasta. From 03.taxonomy onwards every rule reads that file and nothing else about the assembly, so the whole downstream half exists in one copy rather than four.

Terms used on this page

ONT	Oxford Nanopore Technologies — the long-read sequencing platform
QC	quality control
AMR	antimicrobial resistance
IS	insertion sequence — the smallest mobile element, carrying only the genes it needs to move

	illumina	nanopore	hybrid	contigs
Input	paired-end reads	ONT long reads	both, same isolate	a FASTA
Read QC	PhiX removal (Bowtie2), fastp	NanoPlot, Filtlong	both	—
Assembler	SPAdes --isolate	Flye	Flye, plus SPAdes for the comparator	—
After assembly	length and coverage filter	dnaapler, then Medaka*	dnaapler, Medaka*, then Polypolish	header-aware filter
Screened	the SPAdes draft	the reoriented Flye assembly	the Illumina draft; the ONT leg is screened at read level	the supplied contigs
Delivered genome	the screened contigs	the polished assembly	the Polypolish-corrected ONT assembly	the screened contigs
Front-end rules	6	7	16	1

* optional: parameters.{nanopore,hybrid}.medaka_model: false skips the polishing step, and the rule is then not defined at all.

What is identical in all four

Taxonomy, annotation, AMR and virulence screening, plasmid and prophage detection, the optional mobilome module and the MultiQC report are one set of rules, run over contigs_final.fasta. So are the contamination screen and assembly QC — those two sit inside the front end, at the point each mode needs them, but they are the same code everywhere. Only the mapper that builds the coverage track differs: Bowtie2 where there are short reads, minimap2 otherwise.

That is what makes a cross-technology comparison meaningful: when two answers come out of two sets of rules, every difference between them is confounded by the code. Out of one set, the difference is the data. See Rationale.

What the mode still decides downstream

A few shared steps need something a mode does not have. They are gated on the capability rather than on the mode name — HAS_SHORT_READS, HAS_LONG_READS, HAS_READS in workflow/rules/shared/00_common.smk — and when the capability is absent the rule is never defined, so nothing waits for it.

Shared step	Needs	illumina	nanopore	hybrid	contigs
CARD read mapping (05.amr/mapping)	short reads	✓	—	✓	—
Qualimap mapping evaluation	any reads	✓	✓	✓	—
Coverage that carries information for BlobTools	any reads	✓	✓	✓	—
Circular-replicon table handed to Bakta	Flye's topology calls	—	✓	✓	—
Second BLAST of the final assembly, for the plasmid stage	long reads	—	✓	✓	—
IS copy number from reads (mobilome, optional)	short reads	✓	—	✓	—
Two genomes through QC and taxonomy, plus Snippy	both technologies	—	—	✓	—

The four front ends

Page	What it covers
Illumina	PhiX removal, fastp, SPAdes, and the length and coverage cutoffs
Nanopore	NanoPlot, Filtlong, Flye, dnaapler, Medaka — and the filter settings that decide whether a small plasmid survives
Hybrid	how the Illumina leg becomes the ONT leg's contamination filter, Polypolish, and the Snippy comparison
Pre-assembled contigs	what runs when the input is a finished genome, and why the coverage filter is conditional

Choosing a mode answers the other question: which one a given dataset belongs in.