Output files

This page lists the files and folders a FunFlux run writes, and whether each one is still there when the run has finished. Every path on this page is relative to directories.output_dir, and a relative value of that key is read from the launch folder, the folder Snakemake is started from. The numbered folders are the same in every mode, and a mode with nothing to put into a folder does not create it.

Most tables name the rule that writes each file, as the console and logs/ name it. They also give the file's status, which says whether it stays, and the modes or settings under which it is written. The terms below define the labels of the Status and Modes columns.

Terms used on this page

deliverable	a result to read, keep or pass on; a later step may read it too
intermediate	a working file that one step hands to the next; kept after the run
temporary	deleted by Snakemake once the last step that reads it has finished; absent from a finished run
scratch	deleted by the step that wrote it, when that step ends; a download whose checksum did not match is kept instead, for you to inspect
cleanup	deleted by cleanup, the last step of each sample
screen	the contamination and mitochondrion screen, always run in illumina, longread and hybrid mode, and in contigs mode only with parameters.contigs.screen: true
SignalP, GeneMark, Tiberius	for SignalP, parameters.signalp6.mode is not off; for GeneMark, GeneMark-ES is enabled (directories.genemark_dir set, parameters.genemark.mode: auto); for Tiberius, parameters.tiberius.mode is on, the default
AMR	antimicrobial resistance; as a label, the AMR screen is switched on: parameters.amr.chroquetas is on (off by default)

Files to read first

These are the files that hold the results of a run, with the question each one answers and the modes that write it.

File	What it answers	Modes
02.assembly/{sample}/contigs_final.fasta	The nuclear genome. QUAST, BUSCO, the ITS search and funannotate read this file and no other version of the assembly.	all
02.assembly/{sample}/contigs_mitochondrial.fasta	The confirmed mitochondrial contig(s); an empty file when none was confirmed.	screen
02.assembly/{sample}/mitochondria/mitochondrial_screen.tsv	Why each contig was or was not called mitochondrial.	screen
02.assembly/{sample}/contaminants/contig_taxonomy_decisions.tsv	Genus, keep or remove, and the reason, for every contig.	screen
02.assembly/{sample}/eval/busco/short_summary.specific.<lineage>.{sample}.txt	Completeness of the genome.	all
02.assembly/{sample}/eval/quast/report.tsv	Contig count, total length, N50, GC content.	all
02.assembly/{sample}/eval/contig_summary.tsv	What each contig of the draft assembly is (chromosome, chromosome arm, mitochondrion, rDNA array, low-depth piece, collapsed repeat, nuclear contig or fragment, contaminant, another organism's mitochondrial DNA, unclassified), with the evidence; it also links each contig to its scaffold_N name in the annotation.	all
03.taxonomy/{sample}/{sample}_ITS1_taxonomy.txt, {sample}_ITS2_taxonomy.txt	UNITE classification of the ITS barcode.	all
03.taxonomy/{sample}/{sample}_ITS_copies.txt	How many rDNA units were found, and which records the length guard rejected.	all
04.annotation/{sample}/funannotate/annotate_results/	The annotated genome: GenBank, GFF3, proteins, transcripts, and .annotations.txt, the functional annotation with one row per transcript.	all
04.annotation/{sample}/contigs_sort.fasta	The genome as annotated.	all
05.amr/{sample}/amr_summary.tsv	The FungAMR mutations found in the genome, one row per target protein of the species, with a verdict on resistance and the drugs with evidence for and against it.	AMR
06.report/multiqc_report.html	QUAST and BUSCO of every sample of the run on one page; in the read-based modes also the read QC (fastp, NanoPlot) and the mapping QC (Qualimap).	all

The annotation and the assembly use different contig names

funannotate sort renames the contigs scaffold_1, scaffold_2, ..., longest first, before repeat masking, so the masked genome and every annotation file use these names. The screen, QUAST, BUSCO and the ITS files use the names of contigs_final.fasta. The annotated_as column of 02.assembly/{sample}/eval/contig_summary.tsv links the two, as described on Contig summary, and 04.annotation/{sample}/contigs_clean.fasta holds the same sequences under their original names.

Layout

The tree shows the folders of an output folder and the main files in them. A label in square brackets names the modes that write a line, or the setting it needs. On the eval/ line, the label covers only qualimap/. A line without a label is written in every mode, although what 00.resources/ holds depends on the mode and the settings, as its table shows.

<output_dir>/
├── 00.resources/                        run-level material: databases, Tiberius model, SignalP 6 installation, ChroQueTas, a named BUSCO lineage
├── 01.reads/{sample}/                   [illumina, longread, hybrid]
├── 02.assembly/{sample}/
│   ├── contigs_final.fasta              nuclear genome
│   ├── contigs_all.fasta                screened genome, mitochondrion included   [screen]
│   ├── contigs_mitochondrial.fasta                                                [screen]
│   ├── contigs_filt.fasta               the draft the screen examines   [contigs + screen, illumina]
│   ├── contaminants/                    BLAST, BlobTools, keep/remove decisions   [screen]
│   ├── mitochondria/                    mitochondrion screen                      [screen]
│   ├── spades/                                                                    [illumina]
│   ├── hifiasm/ or flye/, assembly.fasta, assembly_evidence.tsv, assembly_provenance.tsv   [longread, hybrid]
│   ├── {sample}_medaka_model.txt, medaka/            [longread, hybrid: ONT samples with Medaka]
│   ├── polypolish/, guide/                                                        [hybrid]
│   └── eval/                            quast/, busco/, telomere_*, adapter_*, contig_summary*.tsv; qualimap/ [illumina, longread, hybrid]
├── 03.taxonomy/{sample}/                ITS windows, ITS sequences, UNITE classification
├── 04.annotation/{sample}/              masking, gene prediction, functional annotation
├── 05.amr/{sample}/                     antifungal resistance mutations (ChroQueTas)   [AMR]
├── 06.report/                           MultiQC
└── logs/                                one log per rule and sample

Disk use

One sample in contigs mode takes in the order of 10 GB. The table shows where the space goes.

Folder	What takes the space	Size
01.reads/{sample}/, 02.assembly/{sample}/ (read-based modes only)	the kept reads in 01.reads/ (the filtered long reads, uncompressed; in hybrid also the selected pairs of Illumina reads), the assembler's working folder (spades/, hifiasm/ or flye/), the read-mapping BAM (two in hybrid) and the polishing output	3.7 to 30 GB for the two folders together in the runs below
04.annotation/{sample}/	most of the rest: the InterProScan XML, there twice (iprscan/interproscan_out.xml and funannotate's copy funannotate/annotate_misc/iprscan.xml), and funannotate's working files in funannotate/predict_misc/ and annotate_misc/	the XML can be a few GB
02.assembly/{sample}/	the BUSCO result tree	0.65 to 1.8 GB with the automatic lineage in the runs below
00.resources/	the antiSMASH 8.0.4 database, unless directories.antismash_db is set	9.4 GB
00.resources/	the SignalP 6 package, with the fast model	1.6 GB
00.resources/	UNITE: the reference and, by default, its copy of the records that name a genus	354 MB (207 + 147 MB)
00.resources/	the Tiberius model	31 MB
00.resources/	the BUSCO lineage data, when parameters.busco.lineage names a lineage	a few hundred MB
00.resources/	ChroQueTas with its FungAMR database (AMR screen on)	3 MB
00.resources/	the telomere repeat database	0.6 MB

In the long-read modes, the kept long reads are the largest single item. Apart from them, the annotation and the assembly folder take most of a sample's space. The first keeps the InterProScan XML twice, the second the read-mapping BAM and the assembler's working files. Of the material shared by the run, the antiSMASH database is the largest item. With directories.antismash_db set, 00.resources/ holds only links to your copy.

To free space once you have checked the results, run workflow/scripts/clean_funflux_output.sh <output folder>. The script lists the bulky working files that no result needs, and deletes them only with --run:

• assembler caches and read mappings;
• the working files and copies of Medaka, BUSCO and funannotate;
• the antiSMASH and UNITE databases and the BUSCO lineage data FunFlux downloaded;
• the licensed SignalP 6 package.

The reads kept in 01.reads/ go only with --include-reads. The results, the logs and tbl2asn's validation reports stay.

Cleaning is for a folder that FunFlux will not use again

The script refuses a folder in which a sample has not finished, but it cannot tell whether another FunFlux run is using the folder. Do not give a cleaned folder to FunFlux again: the deleted files would be made again and most of each sample repeated.

The conda environments are not counted here. They are in .snakemake/ of the launch folder, or in the --conda-prefix folder, as listed under Outside the output folder.

Disk use per run

The sizes below were measured on two public genomes, with one sample per run, and are given as du -sh prints them. The runs are described on Validation on public genomes. The four B05.10 runs with ONT reads were made with FunFlux 2.1.0, the others with FunFlux 2.0.0.

Genome	Run	01.reads/	02.assembly/	Whole output folder
Botrytis cinerea B05.10, 42.6 Mb	illumina, Illumina 79x	0.6 MB	3.7 GB	13 GB
	longread, ONT 42x, hifiasm, Medaka	3.1 GB	7.1 GB	20 GB
	longread, ONT 136x, hifiasm	9.8 GB	12 GB	31 GB
	longread, ONT 136x, Flye	9.8 GB	6.6 GB	26 GB
	hybrid, ONT 136x plus Illumina 79x, hifiasm	16 GB	14 GB	38 GB
Cordyceps cateniannulata GXU-8616, 32.7 Mb	longread, PacBio HiFi 85x, hifiasm	4.7 GB	3.1 GB	17 GB
	longread, PacBio HiFi 85x, Flye	4.7 GB	2.3 GB	16 GB
	hybrid, PacBio HiFi 85x plus MGISEQ short reads 151x, hifiasm	14 GB	5.7 GB	29 GB

A whole sample took 13 to 38 GB in these runs, and the kept long reads were the largest single file of every run with long reads. Their size follows the bases sequenced: 9.7 GB from the 5.8 Gb of ONT reads of B05.10, 4.7 GB from the 2.8 Gb of HiFi reads of GXU-8616.

• In hybrid, 01.reads/ also holds the selected pairs, uncompressed: 9.3 GB of the 14 GB for GXU-8616.
• 02.assembly/ was smaller with Flye than with hifiasm on both genomes, because hifiasm keeps larger working files: hifiasm/ took 5.5 GB on B05.10 and 1.5 GB on GXU-8616, flye/ 860 MB and 570 MB.
• The whole output folder of every run includes about 2 GB of 00.resources/, most of it the SignalP 6 package. It does not include the antiSMASH database, because these runs set directories.antismash_db.

00.resources

00.resources/ holds the material that belongs to the run, not to a sample. It is kept, so a later run on the same output folder downloads or installs none of it again unless the settings behind it change.

Path	Content	Written by	Status	Modes
unite/unite_its_sintax.fasta	UNITE reference for SINTAX, unpacked from links.unite_its_link (207 MB, 266,589 sequences with the default link); with directories.unite_db, a symbolic link to your copy	unite_db or unite_db_local	intermediate	all
unite/PROVENANCE.txt	source URL (or source folder), UTC date, md5 and size in bytes of that FASTA	same	deliverable	all
unite/unite_its_sintax.identified.fasta	the records of that FASTA whose taxonomy names a genus (147 MB, 180,343 sequences with the default link; 86,246 are dropped); what SINTAX classifies against while parameters.its.unite_identified_only is true	unite_identified	intermediate	all
unite/identified_filter.txt	two lines: kept_with_genus: <n> and dropped_without_genus: <n>, the records the filter kept and dropped (180,343 and 86,246 with the default link)	unite_identified	deliverable	all
antismash/databases/	antiSMASH database from download-antismash-databases; with directories.antismash_db, one symbolic link per entry of your copy	antismash_db or antismash_db_local	intermediate	all
funannotate_db_paths.normalized.json	marker that the paths inside the funannotate database were repaired, with the number of files checked and paths rewritten	normalize_funannotate_db	intermediate	all
signalp6/installation.json	archive name and SHA-256, SignalP version, mode, model folder, and the conda environment SignalP 6 was installed into	install_signalp6	intermediate	SignalP
signalp6/package/	the SignalP 6 package with the model weights of the DTU archive (1.6 GB, fast model)	install_signalp6	intermediate	SignalP
signalp6/.matplotlib/	matplotlib cache	install_signalp6	intermediate	SignalP
signalp6/.signalp6-install-*/	the unpacked archive during installation	install_signalp6	scratch	SignalP
tiberius/model/	the Tiberius fungi model, unpacked from links.tiberius_model_link after its SHA-256 matched parameters.tiberius.model_sha256: model_config.json, model_layers.json, weights.h5 (31 MB)	tiberius_model	intermediate	Tiberius
tiberius/model.tar.gz	the downloaded archive; deleted once unpacked, kept when its checksum does not match	tiberius_model	scratch	Tiberius
telomeres/telo_database_v2.tsv	the telomere repeat database of tidk, downloaded from links.telomere_db_link after its SHA-256 matched parameters.contig_summary.telomere_db_sha256: 4,051 rows, one per examined species, with the telomere repeat unit and its support (0.6 MB)	telomere_db	intermediate	all
telomeres/PROVENANCE.txt	source URL, download date, SHA-256 and row count of that file	telomere_db	deliverable	all
telomeres/telo_database_v2.tsv.failed	the downloaded file when its checksum did not match; absent otherwise	telomere_db	scratch	all
adapters/univec_illumina.fasta	the 151 Illumina adapter and primer records of NCBI UniVec build 10.1, downloaded from links.univec_link after its SHA-256 matched parameters.contig_summary.univec_sha256; the contig adapter screen searches for them	univec_illumina	intermediate	all
adapters/PROVENANCE.txt	source URL, download date, SHA-256 and the number of Illumina records kept	univec_illumina	deliverable	all
adapters/univec_illumina.fasta.download.failed	the downloaded file when its checksum did not match; absent otherwise	univec_illumina	scratch	all
chroquetas/ChroQueTas.sh	the ChroQueTas script, version 1.0.1, downloaded unchanged from links.chroquetas_link after its SHA-256 matched parameters.amr.chroquetas_script_sha256 (29 kB)	chroquetas_db	intermediate	AMR
chroquetas/FungAMR_db/	the FungAMR database of ChroQueTas: one folder per species (57), each with a <target>.faa (the reference protein) and a <target>.txt (its resistance positions) per target; unpacked from db-v20250811.tgz after its SHA-256 matched parameters.amr.chroquetas_db_sha256	chroquetas_db	intermediate	AMR
chroquetas/db-v20250811.tgz	the downloaded database archive; deleted once unpacked	chroquetas_db	scratch	AMR
chroquetas/species.txt	the database's species folders, one per line, sorted	chroquetas_db	intermediate	AMR
chroquetas/LICENSE	the licence of ChroQueTas and its database, CC BY-NC-ND 4.0	chroquetas_db	deliverable	AMR
chroquetas/PROVENANCE.txt	the three source addresses, the UTC date, the checksums, the species count and the licence	chroquetas_db	deliverable	AMR
chroquetas/ChroQueTas.sh.failed, chroquetas/db-v20250811.tgz.failed	a download whose checksum did not match; absent otherwise	chroquetas_db	scratch	AMR
phix/	phix.fna.gz, the PhiX genome from links.phix_link, still compressed, and its six bowtie2 index files phix.*.bt2	download_phix, build_phix	intermediate	illumina, hybrid
busco/lineages/<lineage>/	the BUSCO dataset named in parameters.busco.lineage, downloaded once and read by every BUSCO job with --offline (a few hundred MB); later runs into the folder use the same release	busco_data	intermediate	all, with a named lineage
busco/file_versions.tsv	BUSCO's list of the dataset versions it downloaded	busco_data	deliverable	all, with a named lineage

The folders named in directories.unite_db and directories.antismash_db are only read. signalp6/package/ is licensed DTU material, so remove it from any copy of the output folder you share, as explained on Licensing. chroquetas/ is CC BY-NC-ND 4.0 material: it may be passed on unchanged, for non-commercial purposes and with credit.

01.reads

01.reads/ holds the processed reads and their QC reports, in one folder per sample and read type. It is not created in contigs mode.

Path	Content	Written by	Status	Modes
{sample}/illumina/{sample}_nophix_R1.fastq, _R2.fastq	the read pairs that did not align to the PhiX genome, uncompressed	map_phix	temporary	illumina, hybrid
{sample}/illumina/{sample}_trim_R1.fastq, _R2.fastq	the PhiX-free pairs after fastp: adapters and low-quality ends cut, reads under 75 bp dropped	trim_adapters	temporary	illumina, hybrid
{sample}/illumina/{sample}_fastp.html, {sample}_fastp.json	fastp's read-QC report as a page and as numbers; MultiQC reads the JSON	trim_adapters	deliverable	illumina, hybrid
{sample}/long/adapters/{sample}_adapter_scan.tsv	ONT samples: the reads that start with the ONT adapter, and those with a barcode block and no adapter, in every block of 10,000 reads	find_ont_adapter	deliverable	longread, hybrid
{sample}/long/adapters/{sample}_adapter_choice.tsv	ONT samples: what fastplong removes and why, with the numbers behind the decision and ONT's published barcodes found in the reads	find_ont_adapter	deliverable	longread, hybrid
{sample}/long/adapters/{sample}_fastplong_options.txt, {sample}_adapter_sequence.fasta	ONT samples: the options handed to fastplong, and every trimmed sequence in both orientations as FASTA (empty when nothing was trimmed); fastplong reads the FASTA with -a, and the contig adapter screen searches for those sequences too	find_ont_adapter	intermediate	longread, hybrid
{sample}/long/{sample}_trimmed.fastq.gz	ONT samples: the reads after fastplong; Filtlong reads them	trim_ont_adapters	temporary	longread, hybrid
{sample}/long/adapters/{sample}_fastplong.html, {sample}_fastplong.json	ONT samples: fastplong's report, as a page and as numbers (reads trimmed, reads split)	trim_ont_adapters	deliverable	longread, hybrid
{sample}/long/adapters/{sample}_adapter_check.tsv	ONT samples: the trimmed reads that still start with the adapter	check_ont_trimming	deliverable	longread, hybrid
{sample}/long/{sample}_filtered.fastq	the long reads Filtlong kept, uncompressed; the assembler, the coverage mapping and Medaka read it	filter_long_reads	intermediate	longread, hybrid
{sample}/long/raw_qc/, filtered_qc/	NanoPlot on the raw and on the filtered reads: {sample}_NanoStats.txt (MultiQC reads it), an HTML report, the plots as HTML pages (no PNG, --no_static)	raw_long_read_qc, filtered_long_read_qc	deliverable	longread, hybrid
{sample}/selected/{sample}_selected_R1.fastq, _R2.fastq	the trimmed Illumina pairs properly paired on the screened guide assembly; Polypolish reads them, and so does Filtlong for ONT samples with short_read_guidance: true	map_sel_contigs	intermediate	hybrid

02.assembly

02.assembly/ holds one folder per sample with the genome files, the draft and the assembler's files, the results of the screen and the assembly QC. In contigs mode without the screen, the folder holds only contigs_final.fasta and eval/.

Path	Content	Written by	Status	Modes
contigs_final.fasta	the nuclear genome	filter_contigs (no screen) or separate_mitochondria	deliverable	all
contigs_all.fasta	the screened genome, mitochondrion included	finalize_assembly	deliverable	screen
contigs_mitochondrial.fasta	the confirmed mitochondrial contigs, headers unchanged; an empty file when none was confirmed or parameters.mitochondria.mode is off or report. A contig whose best mitochondrial BLAST hit is an animal's or a plant's, such as a contig of human mitochondrial DNA, is classed foreign_mitochondrial and is not confirmed: it stays in the assembly, where the contamination screen keeps or removes it like any other contig	separate_mitochondria	deliverable	screen
contigs_filt.fasta	the draft the screen examines: your FASTA after the contig filter (contigs), or the SPAdes contigs of at least 500 bp and 2x coverage (illumina)	filter_contigs	intermediate	contigs + screen, illumina
spades/	SPAdes' working folder: the unfiltered contigs.fasta, scaffolds.fasta, the assembly graph, one K<k>/ folder per k-mer size, spades.log; emptied whenever SPAdes runs again	illumina_assembly	intermediate	illumina
hifiasm/ or flye/	the assembler's own files: hifiasm's graphs (hifiasm.p_ctg.gfa, the primary contigs; hifiasm.a_ctg.gfa) and .bin files, or Flye's assembly.fasta, assembly_info.txt, graph and working files; emptied whenever the assembler runs again	long_read_assembly	intermediate	longread, hybrid
assembly.fasta	the long-read draft the screen examines, one line per contig (hifiasm: the primary contigs; Flye: a copy of flye/assembly.fasta)	long_read_assembly	intermediate	longread, hybrid
assembly_evidence.tsv	per contig: contig, source, circular, coverage, repeat, alt_group, taken from hifiasm/hifiasm.p_ctg.gfa or flye/assembly_info.txt; the mitochondrion screen and the contig summary read it	assembly_evidence	deliverable	longread, hybrid
assembly_provenance.tsv	field and value columns: assembler, version, read type, input file, command line	long_read_assembly	deliverable	longread, hybrid
{sample}_medaka_model.txt	one line: the Medaka model name after the check, or the path of the model file auto found; the assembler waits for it, Medaka reads it	check_medaka_model	intermediate	longread, hybrid: ONT samples with Medaka
medaka/consensus.fasta	the screened draft polished by Medaka; medaka/ also holds Medaka's working files and is emptied whenever Medaka runs again	long_read_consensus	intermediate	longread, hybrid: ONT samples with Medaka
polypolish/polished.fasta	the draft after correction with the selected pairs, headers cut at the first blank, one line per sequence	short_read_correction	intermediate	hybrid
guide/	the Illumina guide assembly and its screen (below)			hybrid

What the contig filter, rule filter_contigs, does depends on the mode and, in contigs mode, on the first FASTA header:

Mode and draft	What filter_contigs does
contigs, SPAdes-style headers (>NODE_12_length_45678_cov_23.4)	the SPAdes filter: keeps contigs of at least 500 bp and 2x coverage
contigs, any other header	cuts each header at its first space or tab; removes nothing
illumina, SPAdes' contigs.fasta	the same SPAdes filter; an empty result stops the run

An assembly from another assembler therefore loses no contig at this step, only the text after the first space or tab of its headers. See Pre-assembled contigs.

contaminants/

contaminants/ holds the contamination part of the screen: the BLAST hits, the BlobTools table, the decisions of the contig selector (select_contigs) and the coverage track. In the read-based modes it also holds the read depth of every contig, measured from the coverage track, for the contig summary.

Path	Content	Written by	Status
blastout	megablast hits of every contig, 15 tab-separated columns	blast_contigs	intermediate
bestscore.blob.blobDB.table.txt	BlobTools table: per contig, length, GC, coverage and the best-sum taxon at every rank	blob_table	deliverable
contig_taxonomy_decisions.tsv	contig, assigned_genus, action (keep, remove), reason	select_contigs	deliverable
{sample}_composition.txt	share of the DNA and of the contigs per genus, over every contig, kept or not	select_contigs	deliverable
contigs.list	IDs of the kept contigs	select_contigs	deliverable
contigs_decontam.fasta	the kept contigs	select_contigs	intermediate
{sample}_map.bam, {sample}_map.bam.bai	the coverage track: the trimmed Illumina pairs (illumina, bowtie2) or the filtered long reads (longread, hybrid, minimap2) aligned to the draft; in contigs mode the contigs mapped to themselves, only because BlobTools needs a BAM, and the coverage BlobTools derives from it means nothing	map_contigs	kept in the read-based modes (Qualimap and contig_depth read it too); temporary in contigs mode
{sample}_depth.tsv	the read depth of every contig, measured by samtools coverage from {sample}_map.bam: #rname (the contig), startpos, endpos, numreads, covbases, coverage (the percentage of the contig covered), meandepth (the mean read depth), meanbaseq, meanmapq. The depth in the contig summary is meandepth; FunFlux 2.0.0 took it from the coverage column of the BlobTools table, an estimate that can be several times too low with long reads. Not made in contigs mode, where no depth is measured	contig_depth	intermediate; contig_summary reads it
contigs.*.bt2 (six files)	the bowtie2 index of the draft	index_contigs	temporary (illumina)
blob.blobDB.json, blob.{sample}_map.bam.cov	BlobTools database and coverage file	blob_json	temporary

With the default decontamination.mode: off, every contig is kept and still gets a row in contig_taxonomy_decisions.tsv. The decontamination policy and the meaning of each reason are explained on Decontamination.

mitochondria/

mitochondria/ holds the result of the mitochondrion screen.

Path	Content	Written by	Status
mitochondrial_screen.tsv	one row per contig: class, reason, best organelle hit and its query cover, the assembler columns, concatemer_flag, and best_mito_kingdom (the kingdom of the organism of the best hit: Fungi, Metazoa or Viridiplantae; empty when that cannot be told)	screen_mitochondria	deliverable
confirmed_mitochondrial.ids	the confirmed contig IDs, one per line; empty unless mitochondria.mode: separate_confirmed	screen_mitochondria	intermediate

The columns, classes and reasons are explained on Mitochondria.

eval/

eval/ holds the assembly QC: QUAST, BUSCO and, in the read-based modes, Qualimap, together with the telomere search, the adapter screen and the contig summary.

Path	Content	Written by	Status	Modes
quast/	report.tsv, report.txt, report.html, report.pdf, report.tex, transposed_report.*, basic_stats/, quast.log	genome_assembly_evaluation	deliverable	all
busco/short_summary.specific.<lineage>.{sample}.txt, .json	BUSCO result for the most specific lineage	completeness	deliverable	all
busco/short_summary.generic.<lineage>.{sample}.txt, .json	BUSCO result for the broad lineage; only with busco.lineage: auto	completeness	deliverable	all
busco/run_<lineage>/	full_table.tsv (every expected gene: complete, duplicated, fragmented or missing, and where), missing_busco_list.tsv	completeness	deliverable	all
busco/auto_lineage/, busco/logs/, busco/tmp/	BUSCO's lineage-placement runs, its logs, its tmp/ folder	completeness	intermediate	all
busco/busco_tmp/	the lineage datasets downloaded for this sample; only with busco.lineage: auto	completeness	temporary	all
hmmer_output/, busco_sequences/ in the BUSCO tree	per-gene search results and gene sequences	completeness	scratch	all
qualimap/	Qualimap on contaminants/{sample}_map.bam, the sample's reads on the draft: qualimapReport.html, genome_results.txt (mapping rate, mean depth per contig), raw_data_qualimapReport/; an empty folder when Qualimap failed	map_evaluation	deliverable	illumina, longread, hybrid
telomere_candidates.tsv	every telomere repeat unit weighed for the assembly, the units found at the contig ends and those of the tidk database and of the literature: copies, contig ends in the telomere orientation and in the opposite one, score, whether the unit is known, and which one was chosen	telomere_units	deliverable	all
telomere_unit.txt	one line: the telomere repeat unit of the assembly in canonical form, or none	telomere_units	intermediate; contig_summary reads it	all
telomere_ends.tsv	per contig end, tidk's call for the chosen unit: status, the array's coordinates and length, its distance from the end, its strand	telomere_units	deliverable; contig_summary reads it	all
adapter_query.fasta	the adapter and barcode sequences the screen searches for	blast_adapters	temporary	all
adapter_blast.tsv	BLAST's hits of the adapter and barcode sequences on the draft contigs	blast_adapters	intermediate	all
adapter_screen.tsv	one row per adapter locus on the draft contigs: contig, contig length, start and end of the locus, its distance to the nearer contig end, end or internal, the longest hit and its identity, and the adapters that matched; the header line alone when none was found	screen_adapters	deliverable; contig_summary reads it	all
adapter_screen_mqc.tsv	the sample's row of the Adapters table of the report	screen_adapters	deliverable; MultiQC reads it	all
contig_summary.tsv	one row per contig of the draft assembly: length, GC, read depth and depth relative to the median, the assembler's depth and circularity, the telomere array at each end (copies) and its distance from the end, the telomere unit, rDNA units, BUSCO markers, genes, genus, the screens' calls, the adapter loci, the scaffold_N name, the verdict and its reason; written after the annotation	contig_summary	deliverable	all
contig_summary_mqc.tsv	the same table behind MultiQC's header lines, rows named {sample}: {contig}	contig_summary	deliverable; MultiQC reads it	all

The columns, the verdicts and how the telomere unit is chosen are explained on Contig summary.

guide/ (hybrid)

In hybrid mode, guide/ holds the Illumina guide assembly and its own screen, laid out like the sample folder itself. The guide is not delivered. QUAST reports its screened version under the label {sample}_guide, and the Illumina pairs that map to it are the selected pairs in 01.reads/.

Path	Content	Written by	Status
guide/spades/	SPAdes' working folder for the guide, as spades/ above	illumina_assembly	intermediate
guide/contigs_filt.fasta	the guide after the 500 bp / 2x filter	filter_contigs	intermediate
guide/contaminants/{sample}_map.bam, .bai	the trimmed Illumina pairs aligned to the guide; Qualimap reads it too	map_guide_contigs	intermediate
guide/contaminants/filtered.*.bt2, selected.*.bt2	the bowtie2 indexes of the filtered and of the screened guide	index_guide_contigs, index_selected_contigs	temporary
guide/contaminants/blastout, bestscore.blob.blobDB.table.txt, contig_taxonomy_decisions.tsv, {sample}_composition.txt, contigs.list	the same files as in contaminants/ above, for the guide	blast_guide_contigs, blob_guide_table, select_guide_contigs	as in contaminants/
guide/contaminants/contigs_decontam.fasta	the screened guide: QUAST reports it as {sample}_guide, and the read selection maps to it	select_guide_contigs	intermediate
guide/contaminants/blob.blobDB.json, blob.{sample}_map.bam.cov, selected_map.sam	BlobTools' working files; the alignments of the read selection	blob_guide_json, map_sel_contigs	temporary
guide/mitochondria/mitochondrial_screen.tsv, confirmed_mitochondrial.ids	the mitochondrion screen of the guide, from BLAST alone	screen_guide_mitochondria	deliverable, intermediate
guide/eval/qualimap/	Qualimap on the Illumina pairs mapped to the guide (MultiQC row mapping Illumina QC)	map_guide_evaluation	deliverable

03.taxonomy

03.taxonomy/ holds the ITS taxonomy, in one folder per sample. The method is described on ITS taxonomy.

Path	Content	Written by	Status
its_windows.fasta	one window per rDNA unit, its.window_flank bp (2,000) on each side of the 5.8S gene, turned to read SSU-ITS1-5.8S-ITS2-LSU; headers contig:start-end_strand	its_windows	intermediate
its_windows.tsv	window_id, contig, start, end, strand, anchor_hits, best_evalue, window_length	its_windows	deliverable
{sample}.full.fasta, {sample}.ITS1.fasta, {sample}.ITS2.fasta	ITSx output, one record per rDNA unit, named after its window	its_extraction	deliverable
{sample}.summary.txt	ITSx summary	its_extraction	deliverable
{sample}_ITS_copies.txt	one row per distinct sequence: region, sequence_id (- when rejected), length, copies, status (ok, rejected_too_short, rejected_too_long), windows	its_classification	deliverable
{sample}_ITS1_taxonomy.txt, {sample}_ITS2_taxonomy.txt	SINTAX result for the distinct sequences ({sample}_ITS1_1, ...); no header; record name, taxonomy with a confidence per rank, strand, taxonomy kept at its_taxonomy_cutoff	its_classification	deliverable
{sample}.ITSx_tmp/	ITSx working folder	its_extraction	scratch

The full ITS region is listed in the copies table but not classified. A genome without a recognisable 5.8S gene gives an empty windows file, empty ITS and taxonomy files and a note in the logs, and the run continues.

04.annotation

04.annotation/ holds the annotation, in one folder per sample: the cleaned and renamed genome, the repeat masking, the gene prediction and the functional annotation. The methods are described on Repeat masking, Gene prediction and Functional annotation.

Path	Content	Written by	Status	Modes
contigs_clean.fasta	contigs_final.fasta after funannotate clean, which removes contigs under 500 bp and duplicated contigs shorter than the N50 (the rules); original names	funannotate_preprocess	deliverable	all
contigs_sort.fasta	the same contigs sorted longest first and renamed scaffold_N: the genome that is annotated	funannotate_preprocess	deliverable	all
contigs_removed_by_clean.txt	names of the contigs funannotate clean removed; an empty file when none	funannotate_preprocess	deliverable	all
contigs_mask.fasta	contigs_sort.fasta with repeats in lowercase	repeat_masking	intermediate	all
repeatmasking/	RepeatModeler library and RepeatMasker reports; empty with masking_method: tantan; emptied whenever the rule starts	repeat_masking	intermediate	all
genemark_check.txt	date and outcome of the GeneMark-ES test on a 200 kb slice of contigs_mask.fasta	genemark_check	intermediate	GeneMark
tiberius/tiberius.gtf	Tiberius's gene models (GTF): one transcript per gene, with exons, CDS, start and stop codons	tiberius_prediction	intermediate	Tiberius
tiberius/tiberius.evm.gff3	the same models as GFF3 for EVidenceModeler, the file funannotate predict gets with --other_gff	tiberius_gff3	intermediate	Tiberius
augustus_model.txt	what AUGUSTUS does for the sample: the stored AUGUSTUS model it uses as it is, or training on the genome's BUSCO genes (optimised or not); copied to the top of logs/funannotate_prediction_{sample}.log	augustus_model_check	intermediate	all
funannotate/predict_results/	the gene prediction: <Genus_species>_{sample}.gbk, .gff3, .proteins.fa, .cds-transcripts.fa, .mrna-transcripts.fa, .discrepency.report.txt, .error.summary.txt, and <model>.parameters.json (the name of the stored AUGUSTUS model, or <genus>_<species>_<sample> in lower case for a model trained on the genome)	funannotate_prediction	intermediate	all
funannotate/predict_misc/, funannotate/logfiles/	funannotate's working files and logs	funannotate_prediction	intermediate	all
funannotate/annotate_results/	the final annotation: <Genus_species>_{sample}.gbk, .gff3, .tbl, .sqn, .agp, .contigs.fsa, .scaffolds.fa, .proteins.fa, .cds-transcripts.fa, .mrna-transcripts.fa, .annotations.txt, .clusters.txt, .stats.json, .discrepency.report.txt, and Gene2Products.must-fix.txt, Gene2Products.need-curating.txt, Gene2Products.new-names-passed.txt	funannotate_annotation	deliverable	all
funannotate/annotate_misc/	working files of funannotate annotate, among them annotations.secretome.txt (with SignalP), genome.fixedproducts and iprscan.xml, funannotate's copy of the InterProScan XML	funannotate_annotation	intermediate	all
iprscan/interproscan_out.xml	InterProScan matches with GO terms and pathways (can be a few GB)	iprscan_annotation	deliverable	all
iprscan/tmp/	InterProScan's working folder, left in place; empty or nearly so after a finished run	iprscan_annotation	intermediate	all
eggnog/{sample}.emapper.annotations, .emapper.hits, .emapper.seed_orthologs	eggNOG-mapper results	eggnog_annotation	deliverable	all
eggnog/eggnog_tmp/	eggNOG-mapper working folder	eggnog_annotation	temporary	all
antismash/	index.html (the cluster report), antismash.gbk (read by funannotate annotate), antismash.json, antismash.zip, regions.js, one <scaffold>.regionNNN.gbk per region	antismash_annotation	deliverable	all
signalp6/	prediction_results.txt (read by funannotate annotate), output.gff3, output.json, region_output.gff3, processed_entries.fasta, proteins.for_signalp.faa, .matplotlib/	signalp6_annotation	deliverable	SignalP
.signalp6-{sample}.XXXXXX/	SignalP working folder; it replaces signalp6/ only after the result passed its checks	signalp6_annotation	scratch	SignalP
.annotation_complete.txt, .cleanup_complete.txt	empty markers: funannotate annotate finished; the cleanup finished	funannotate_annotation, cleanup	intermediate	all
clean_*/, *.log, genome.fixedproducts directly in this folder	funannotate scratch files, if any are left; listed in logs/cleanup_{sample}.log		cleanup	all

In the file names, <Genus_species> is the species column of the annotation parameter table with spaces turned into underscores, for example Fusarium_graminearum_SAMPLE1.gff3. The sample name is also the locus-tag prefix of every gene ID.

05.amr

05.amr/ is written only when the AMR screen is on, and holds its results in one folder per sample. The method, the target statuses and the columns are explained on Antifungal resistance mutations.

Path	Content	Written by	Status
amr_summary.tsv	one row per target protein of the species: compartment, status (evaluated, not evaluated, not found, not screened), positions compared, the verdict (resistance, conflicting, unknown, no resistance, wild type), the FungAMR mutations with their verdicts and the new mutations found, the drugs with evidence of resistance and of no resistance with their levels, a note	amr_chroquetas	deliverable
amr_positions.tsv	one row per FungAMR position of every copy read position by position: target, compartment, copy, position, reference and found amino acid, call (FungAMR mutation, new mutation, no mutation; or no alignment or position not found where the position could not be compared), verdict, the drugs with evidence of resistance and of no resistance with their levels in words, the changes with evidence of resistance at that position, ChroQueTas's drug text	amr_chroquetas	deliverable
amr_summary_mqc.tsv	the summary behind MultiQC's header lines, rows named {sample}: {target}	amr_chroquetas	deliverable; MultiQC reads it
chroquetas/nuclear/, chroquetas/mitochondrial/	ChroQueTas's own output of the two runs, unchanged: its summary and statistics, one table per target copy, the extracted proteins, miniprot's alignments, the BLASTP comparisons, tmp/. A run that was not made has no folder: mitochondrial/ exists only when the species has cytochrome b and its mitochondrial run was made	amr_chroquetas	deliverable

06.report

06.report/ holds the MultiQC report of the run and the files behind it.

Path	Content	Status
multiqc_report.html	QUAST and BUSCO results, the contig summary table and the Adapters table of every sample of the run, and the AMR table with the AMR screen on; in the read-based modes also fastp (illumina, hybrid), NanoPlot (longread, hybrid) and Qualimap	deliverable
multiqc_data/	the tables behind the plots, for example multiqc_general_stats.txt, multiqc_busco.txt, multiqc_quast.txt, multiqc_contig_summary.txt, multiqc_adapter_screen.txt	deliverable
multiqc_config.yaml	the MultiQC configuration of the run, with the sample-name rules	intermediate

Rows are named after the tool and the sample, for example QUAST assembly QC | <sample> and BUSCO completeness QC (<lineage>) | <sample>. A sample has two BUSCO rows with the automatic lineage and one with a named lineage. The rows of the contig summary table are named {sample}: {contig}, and those of the AMR table {sample}: {target}. The rows of each mode are listed under Report. The report covers the samples of the current run only and is rewritten when they change.

logs

logs/ holds a log for every per-sample rule and sample, and the logs of the steps that run once for the whole run.

File	Written by
<rule>_{sample}.log	every per-sample rule, for example funannotate_prediction_SAMPLE1.log
funannotate_db.log, normalize_funannotate_db.log	the funannotate database download and path repair
unite_db.log	unite_db or unite_db_local
busco_data.log	busco_data: the download of a named BUSCO lineage
unite_identified.log	unite_identified: the one-line summary of the filter and the two counts
telomere_db.log	telomere_db: the download of the telomere repeat database and its checksum
univec_illumina.log	univec_illumina: the download of UniVec, its checksum and the Illumina records kept
chroquetas_db.log	chroquetas_db: the download of ChroQueTas and its database, the two checksums, the unpacking (AMR screen on)
tiberius_model.log	tiberius_model: the download of the Tiberius fungi model, its checksum and the unpacking (Tiberius on)
antismash_db.log or antismash_db_local.log	the antiSMASH database rule of the run
install_signalp6.log	the SignalP 6 installation
download_phix.log, build_phix.log	the PhiX genome download and its bowtie2 index (illumina, hybrid)
multiqc.log	the report

Snakemake deletes the log of a job before it runs that job again, so each log holds the latest attempt only. Copy a log you want to keep before you run again. Some messages go only to the logs:

Only in the log	Log
the warnings of the contig selector	select_contigs_{sample}.log
a QUAST, Qualimap or NanoPlot failure	genome_assembly_evaluation_{sample}.log, map_evaluation_{sample}.log, map_guide_evaluation_{sample}.log, raw_long_read_qc_{sample}.log, filtered_long_read_qc_{sample}.log
a genome without 5.8S	its_extraction_{sample}.log
the share of PhiX in an Illumina library	map_phix_{sample}.log
fastp's and Filtlong's read totals	trim_adapters_{sample}.log, filter_long_reads_{sample}.log
how many SPAdes contigs the filter kept	filter_contigs_{sample}.log
which file became contigs_all.fasta	finalize_assembly_{sample}.log
tidk's messages and the candidate units behind the telomere unit (the [FunFlux] telomere_units summary line and its NOTE reach the console too)	telomere_units_{sample}.log
the one-line summary of the contig summary: contigs per verdict, contig ends with a telomere, median depth, BUSCO lineage	contig_summary_{sample}.log
ChroQueTas's messages of both runs, among them the targets it could not extract (the [FunFlux] amr_chroquetas summary line and its NOTE reach the console too)	amr_chroquetas_{sample}.log

Snakemake writes its own log of each run to .snakemake/log/ in the launch folder. See Troubleshooting.

Temporary files and folders a re-run replaces

The files and folders below are deleted during the run, either by Snakemake once no later step needs them or by the step that made them. A finished run has none of them, so they take space only while the run goes on.

Path	Modes
01.reads/{sample}/illumina/{sample}_nophix_R1.fastq, _R2.fastq	illumina, hybrid
01.reads/{sample}/illumina/{sample}_trim_R1.fastq, _R2.fastq	illumina, hybrid
01.reads/{sample}/long/{sample}_trimmed.fastq.gz	longread, hybrid (ONT samples)
02.assembly/{sample}/eval/adapter_query.fasta	all
02.assembly/{sample}/contaminants/{sample}_map.bam and .bai	contigs + screen (kept in the read-based modes)
02.assembly/{sample}/contaminants/contigs.*.bt2	illumina
02.assembly/{sample}/contaminants/blob.blobDB.json, blob.{sample}_map.bam.cov	screen
02.assembly/{sample}/guide/contaminants/filtered.*.bt2, selected.*.bt2, blob.blobDB.json, blob.{sample}_map.bam.cov, selected_map.sam	hybrid
02.assembly/{sample}/polypolish/draft.fasta, its bwa index files, align_1.sam, align_2.sam, filtered_align_1.sam, filtered_align_2.sam, polished_raw.fasta	hybrid
02.assembly/{sample}/eval/busco/busco_tmp/ (busco.lineage: auto), and BUSCO's hmmer_output/ and busco_sequences/	all
03.taxonomy/{sample}/{sample}.ITSx_tmp/	all
04.annotation/{sample}/eggnog/eggnog_tmp/	all
04.annotation/{sample}/.signalp6-{sample}.XXXXXX/, the SignalP working folder, and 00.resources/signalp6/.signalp6-install-*/	SignalP
a folder in the system's temporary directory, for its_windows and its_classification	all
a folder in the system's temporary directory, for genemark_check	GeneMark

The folders below are outputs that Snakemake deletes before it runs their rule again.

Folder	Rule
01.reads/{sample}/long/raw_qc/, filtered_qc/	raw_long_read_qc, filtered_long_read_qc
02.assembly/{sample}/spades/, 02.assembly/{sample}/guide/spades/	illumina_assembly
02.assembly/{sample}/hifiasm/ or flye/	long_read_assembly
02.assembly/{sample}/medaka/	long_read_consensus
02.assembly/{sample}/eval/quast/	genome_assembly_evaluation
02.assembly/{sample}/eval/busco/	completeness
02.assembly/{sample}/eval/qualimap/, 02.assembly/{sample}/guide/eval/qualimap/	map_evaluation, map_guide_evaluation
04.annotation/{sample}/funannotate/	funannotate_prediction
04.annotation/{sample}/antismash/	antismash_annotation
04.annotation/{sample}/eggnog/	eggnog_annotation
05.amr/{sample}/chroquetas/	amr_chroquetas
00.resources/antismash/databases/	antismash_db or antismash_db_local
06.report/multiqc_data/	multiqc

Files you put into these folders are lost on a re-run

The same holds for repeatmasking/, which repeat_masking empties itself, and for signalp6/, which signalp6_annotation replaces as a whole. Because funannotate annotate writes into funannotate/, a re-run of funannotate_prediction also deletes annotate_results/, and the annotation runs again. With antismash_db_local, a re-run deletes only the links, never your database.

.snakemake_timestamp inside such a folder is Snakemake's time stamp for it, so leave it in place. What triggers a re-run is explained on Running FunFlux.

Outside the output folder

A run also writes outside directories.output_dir, in the places below.

Where	What
directories.funannotate_db	the funannotate database: archive (funannotate_db_1.8.17_2026-10-08.tar.gz by default, 2.1 GB) and its .sha256, both kept after unpacking; the unpacked database, whose stored paths normalize_funannotate_db rewrites in place; 8.1 GB in total
.snakemake/ in the launch folder	Snakemake's records of what was made and how, locks, markers of unfinished jobs, run logs, and the conda environments unless --conda-prefix names another folder
the BlobTools conda environment	nodesDB.txt, built by BlobTools from nodes.dmp and names.dmp on first use
the Medaka conda environment, or ~/.medaka/data when it cannot be written	the weights of each Medaka model, downloaded by the model check the first time a model is used
the pip cache of the user	the wheel pip builds while installing SignalP 6

The databases are described on Reference databases, and the environments and the launch folder on Running FunFlux.