Overview

FunFlux is a Snakemake workflow for fungal isolate genome analysis. It accepts Illumina reads, Oxford Nanopore (ONT) or PacBio HiFi reads, short and long reads from the same isolate, or an existing assembly. The workflow combines assembly, contamination and mitochondrion screening, quality control, ITS-based taxonomy, repeat masking, gene prediction and functional annotation.

FunFlux was developed for the genomic characterisation of fungal isolates. Its results describe the assembled genome, its predicted genes and their functional annotation, including the secondary-metabolite gene clusters found by antiSMASH. They are predictions from the genome sequence, to be weighed with curation and experimental evidence.

An optional screen looks for the point mutations that the FungAMR database lists with evidence for or against resistance to antifungal drugs and fungicides. The configuration key mode selects the input mode. This site documents version v2.1.0.

FunFlux belongs to the BioFlux family of pipelines.

Terms used on this page

ONT	Oxford Nanopore Technologies, a long-read sequencing platform
HiFi	PacBio's accurate long reads, each the consensus of several passes over one molecule (CCS)
ITS	internal transcribed spacer — the rDNA region used as the barcode of fungi
BUSCO	a set of single-copy genes expected in every genome of a lineage, used to estimate completeness
soft-masking	writing repeats in lower case, so that gene finders can treat them differently without losing the sequence
AMR	antimicrobial resistance; in FunFlux, point mutations studied for resistance against antifungal drugs and fungicides

The workflow at a glance

Each coloured line is one mode: follow yours from its input to the report. The four lines take their own routes to contigs_final.fasta, the genome every mode delivers, and from there run on together through the analyses. In hybrid the purple line has two arms: the Illumina arm builds a guide assembly with SPAdes, screens it and keeps the Illumina pairs that map to it, and Polypolish uses those pairs to correct the long-read assembly. In contigs mode the solid gold line goes straight to contigs_final.fasta; the dashed gold path is the contamination and mitochondrion screen, which the configuration can switch on. Where the lines split into parallel tracks, the steps run side by side. Analysis groups after the final genome may run in parallel. The order along the line is a reading order, not the order in which the steps must run. An open ring marks a step the configuration switches on or off. The Medaka model check, fastplong (adapter trimming) and Medaka apply to ONT samples only; HiFi reads pass them by. GeneMark-ES, marked beside funannotate predict, runs inside it when it is switched on. Thin lines are extra inputs, and the dashed red line is the optional resistance-mutation screen. Select the map to open it larger; select it again to see it at full size, and close it with × or Esc.

The stages, their tools and when they run:

Stage	Tools	Runs
Assembly	depends on the mode (below)	illumina, longread, hybrid; contigs mode starts from your assembly
Contamination screen	BLAST against NCBI core_nt, BlobTools	always; in contigs mode only when the screen is switched on
Mitochondrion screen	BLAST, with the assembler's evidence in the long-read modes	with the contamination screen
Assembly QC	QUAST, BUSCO	always
Contig summary	tidk finds the telomeres and BLAST+ screens for adapter sequence; one table per sample gives every contig of the draft a verdict (Contig summary)	always
ITS taxonomy	ITSx, SINTAX against UNITE	always
Repeat masking	tantan, or RepeatModeler and RepeatMasker	always
Gene prediction	funannotate with its own gene finders, Tiberius (on by default) and GeneMark-ES (with a licence)	always
Functional annotation	InterProScan, eggNOG-mapper, antiSMASH, SignalP 6	always; SignalP 6 is optional
Antifungal resistance mutations	ChroQueTas against the FungAMR database	only when switched on
Report	MultiQC, one report for all samples	always

From the contamination screen on, every mode runs the same steps. Only polishing differs between the read-based modes.

The four modes

	illumina	longread	hybrid	contigs
Input	Illumina paired-end reads, {sample}_R1.<ext> and {sample}_R2.<ext>	ONT reads {sample}_ont.<ext> or PacBio HiFi reads {sample}_pacbio.<ext>; one batch may hold both	Illumina pairs and ONT or PacBio HiFi reads for every sample	one assembly per sample, {sample}.fasta, .fa or .fna
Read QC	PhiX removal (bowtie2), then fastp	ONT adapters removed (fastplong); NanoPlot before and after Filtlong	both of the above	none
Assembly	SPAdes	hifiasm (default) or Flye	an Illumina guide assembly (SPAdes) to select the isolate's read pairs, then hifiasm or Flye on the long reads	no assembler
Contig filter	≥ 500 bp and ≥ 2× coverage, read off SPAdes' own headers	none	the same cut-offs, on the guide assembly only	the same cut-offs when the headers are SPAdes-style; otherwise headers are cut at the first space or tab and nothing is removed
Adapter screen	the draft contigs are searched for adapter and barcode sequence (BLAST+); reported, nothing removed	the same	the same	the same
Polishing	none	Medaka, ONT samples*	Medaka (ONT samples*), then Polypolish with the Illumina pairs	none
Delivered genome	the screened SPAdes contigs	the screened long-read assembly, polished with Medaka for ONT samples	the screened long-read assembly, corrected with the Illumina pairs	the input contigs, screened only with parameters.contigs.screen: true

* Medaka is on by default for ONT samples, and medaka_model: false switches it off. PacBio HiFi samples are never polished with Medaka. <ext> is fastq, fq, fastq.gz or fq.gz, the same for all files of one kind in a batch, and _pacbio files must hold HiFi (CCS) reads. CLR reads are not supported.

hybrid requires more computation than either single-technology mode, because it assembles and screens the Illumina reads as well. In hybrid, Filtlong can also score the ONT reads against the selected Illumina pairs, but only when short_read_guidance is switched on. If you used FunFlux 1.x, Funnotator or FunFluxL, Coming from v1 lists what changed. See Modes and Choosing a mode.

Delivered genome files

Every sample ends with up to three FASTA files in 02.assembly/{sample}/. Their content depends on two settings: whether the contamination and mitochondrion screen runs, and parameters.mitochondria.mode. The screen always runs in illumina, longread and hybrid, and in contigs mode only with parameters.contigs.screen: true.

File	Screen on, mitochondrion split off (separate_confirmed, the default)	Screen on, parameters.mitochondria.mode set to off or report	contigs mode without the screen
contigs_all.fasta	the assembly after the contamination screen and, in longread and hybrid, after polishing; mitochondrion included	the same	not written
contigs_final.fasta	the nuclear genome: contigs_all.fasta without the confirmed mitochondrial contigs	every contig	the input after the contig filter, including a mitochondrial genome if the input contains one
contigs_mitochondrial.fasta	the confirmed mitochondrial contigs; an empty file when none was confirmed	an empty file	not written

contigs_final.fasta is the genome that QC, ITS and annotation read

QUAST, BUSCO, the ITS extraction and funannotate run on contigs_final.fasta. It is the nuclear genome alone only when the screen runs with the default separate_confirmed. Otherwise nothing is split off.

The gene models refer to a renamed copy of that genome. funannotate clean and funannotate sort turn contigs_final.fasta into 04.annotation/{sample}/contigs_sort.fasta, whose contigs are named scaffold_1, scaffold_2, ... from the longest down, and the gene models use these names. QUAST, BUSCO and the ITS files keep the names of contigs_final.fasta. contigs_removed_by_clean.txt, next to contigs_sort.fasta, lists the contigs funannotate clean removed.

One command, one configuration, one report

All four modes are launched the same way, from the repository root:

snakemake --sdm conda --cores 24 --configfile config/config_custom.yaml

--sdm conda builds the conda environment of each rule on first use. Besides Snakemake, you obtain InterProScan yourself, and the optional GeneMark-ES and SignalP 6 if you use them (Licensed and external tools). --cores sets the CPU budget of a local run and is used without --jobs. See Running FunFlux.

Pass the configuration file explicitly with --configfile. FunFlux ships a template, config/config.yaml, but does not load it on its own: copy it, then set the mode, the input and output paths, the database locations and the annotation table. A wrong mode, a missing input directory, a bad sample name or a missing annotation row stops the dry run before any job starts. See Configuration and Quick start.

Results are written to numbered folders that are the same in every mode: 00.resources for material shared by the samples of a run, 01.reads (not in contigs mode), 02.assembly, 03.taxonomy, 04.annotation, 05.amr (only with the resistance-mutation screen on), 06.report and logs. The funannotate database stays in directories.funannotate_db, outside the output folder. Every run ends with a MultiQC report, 06.report/multiqc_report.html, which collects the QC of all samples. The annotation itself is in 04.annotation/{sample}/funannotate/annotate_results/. See Output files.

What has to be prepared in advance

FunFlux installs its own software, but three resources must be on disk before the first run: NCBI core_nt with its taxonomy files (not read in contigs mode without the screen), the eggNOG-mapper data, and an InterProScan installation. GeneMark-ES and SignalP 6 are optional and need a licence. Everything else is downloaded on first use, or can be pointed at a copy you already hold. See Reference databases and Licensed and external tools.

Where the resistance-mutation screen fits

Stage 05.amr is disabled by default (parameters.amr.chroquetas: off). When enabled, ChroQueTas searches each genome for the point mutations that the FungAMR database lists with evidence for or against resistance to antifungal drugs and fungicides, and FunFlux summarises for each sample the known changes found and the drugs FungAMR links them to. A sample of a species that FungAMR does not cover is reported as not screened.

These calls describe known mutations in the genes and species that FungAMR covers. They do not establish a resistance phenotype, and the absence of a call is not evidence of susceptibility. Interpret them together with the evidence levels FungAMR gives each mutation. See Antifungal resistance mutations.

Where to go next

Page	Read it for
Choosing a mode	which mode a dataset belongs in, and how to name the input files
Installation	Snakemake (9.14.6 or a later 9.x; tested with 9.14.6), the repository and the conda environments
Reference databases	what to download before the first run, and what FunFlux fetches
Licensed and external tools	InterProScan, which every run needs, and the optional GeneMark-ES and SignalP 6
Quick start	the commands from a filled-in configuration to a finished run
Configuration	every configuration key, the modes that read it, its default and allowed values
Running FunFlux	the dry run, CPU cores and threads, resuming an interrupted run, running on a cluster
Troubleshooting	what a [FunFlux] message or a failed step means, and what to do
Validation on public genomes	what each read-based mode produced on Botrytis cinerea B05.10, what PacBio HiFi reads of Cordyceps cateniannulata GXU-8616 gave, and how to repeat the runs

FunFlux is released under the MIT License; the tools and databases it calls carry their own terms, listed on Licensing.