Troubleshooting

This page lists the messages that stop a FunFlux run or leave a result missing, with the cause of each and what to change. It follows a run from start to end: the checks of the configuration before any job starts, the failures of single steps while the workflow runs, and the stops particular to a cluster.

Terms used on this page

dry run	a Snakemake run with -n: FunFlux checks the configuration and the inputs, and Snakemake lists the jobs it plans without running any
rule	one step of the workflow as Snakemake defines it, such as select_contigs; the console and the log files in logs/ use its name
launch folder	the folder Snakemake is started from, normally the FunFlux repository; relative paths are read from it, and .snakemake/ is created there
lock	the record, in .snakemake/, of the files a run in progress reads and writes; a killed run leaves it behind, and --unlock removes it
the screen	the contamination and mitochondrion screen of the draft: BLAST against core_nt, BlobTools, the mitochondrion screen and the contig selector; in contigs mode only with parameters.contigs.screen: true
contig selector	rule select_contigs, the last step of the screen: it keeps or removes each contig by genus, as parameters.decontamination says; the default, off, keeps every contig
annotation parameter table	the tab-separated file named by files.annotation_params, one row per sample: name, species, a protein FASTA of related fungi and, optionally, an AUGUSTUS model
heavy step, long step, light step	on a cluster, heavy steps ask the scheduler for ram_gb and runtime_hours, long steps for runtime_hours only, and light steps for neither

Where to look

What you see, and when it appears, tells you where the cause is and which part of this page to read.

What you see	Where the cause is
A [FunFlux] ... line before Snakemake's Building DAG of jobs..., and the run ends	A configuration check failed and nothing ran. The message names the key. See Stops before any job runs.
A [FunFlux] <step>: ... line after jobs have started, for example [FunFlux] funannotate_db: ...	That step failed or, if it is a report step such as QUAST, it warns and the run goes on. Look the step up under Failures while the workflow runs. Its log is logs/<step>.log or logs/<step>_{sample}.log.
Error in rule <rule>: while the workflow runs	The block names the job's log:, the file that holds the tool's own messages. See Failures while the workflow runs.
Error: Directory cannot be locked. or The files below seem to be incomplete.	A previous run was killed. Directory cannot be locked has a second cause: another run from the same launch folder would write the same files, or one of the two runs would write a file the other reads. Use --unlock only when no other Snakemake process runs from that folder. See Resuming after an interruption.
A finished run with an empty or missing result	Some messages go only to the logs and never stop the run. See Messages that do not stop the run.
A message from Slurm, a job killed for memory or time, SSL certificate problem, or a step that works on one computer but not on the cluster	See On a cluster.

Three kinds of log are kept:

Log of	Where
every per-sample rule	<output_dir>/logs/<rule>_{sample}.log
the rules that run once for all samples	<output_dir>/logs/, without a sample name: funannotate_db.log, normalize_funannotate_db.log, unite_db.log, unite_identified.log, antismash_db.log or antismash_db_local.log, tiberius_model.log, telomere_db.log, univec_illumina.log, busco_data.log (a named BUSCO lineage), download_phix.log, build_phix.log, install_signalp6.log, chroquetas_db.log, multiqc.log
Snakemake itself	.snakemake/log/ in the launch folder, one file per run, named after its start time

--show-failed-logs makes Snakemake print the log of every failed job at the end of the run.

Stops before any job runs

FunFlux checks the whole configuration before any job starts, on every run and every dry run. A failed check prints one [FunFlux] line and exits with status 1. Only the first failed check is reported, so fix it and run -n again. Lines that start with [FunFlux] NOTE: do not stop the run.

Placeholders in the messages below are in angle brackets. An empty cell means that the message itself already says what is wrong or what to change.

Configuration file and mode

These checks concern the configuration file itself and its mode key.

Message	Cause	Fix
[FunFlux] No configuration supplied. Pass one explicitly, e.g.:	No --configfile on the command line. FunFlux has no default configuration.	Add --configfile config/config_custom.yaml.
[FunFlux] mode: this configuration has no 'mode' key. ...	A v1 configuration (FunFlux 1.x, FunFluxL, FunFluxL+), or an empty mode:.	Start from the v2 template. Coming from v1 lists what moved.
[FunFlux] mode: must be one of illumina | longread | hybrid | contigs (got: '<value>')	Another word in mode.	Use one of the four.

Keys and values

These messages name a key that is missing, unknown or in the wrong block, or whose value has the wrong form or is out of range.

Message	Cause	Fix
[FunFlux] <block>: must be a block of 'key: value' lines.	input, directories, files, links, resources, parameters or a block inside parameters holds a single value or a list.	Indent the keys under the block name.
[FunFlux] <section>.<key>: required, but it is missing or empty in the config.	A required entry is empty: directories.output_dir, eggnog_db, funannotate_db, blast_db (when the screen runs), files.iprscan, files.annotation_params, links.funannotate_link, links.phix_link (in illumina and hybrid mode). The template leaves them empty on purpose.	Fill it in.
[FunFlux] <key>: expected true or false (got <value>).	A yes/no key holds another word. parameters.contigs.screen is checked in every mode, although only contigs mode uses it.	Write true or false.
[FunFlux] <key>: expected a whole number (got <value>)., ... must be at least N (got M)., ... must be at most N (got M).	The value is not an integer, or it is out of range.	The allowed values are listed on Configuration.
[FunFlux] <key>: expected a number (got <value>)., ... must be above 0 (got X)., ... must be at most N (got X).	The same for decimal keys, for example min_direct_blast_qcov: 150. nan and inf, and YAML's .nan and .inf, count as not a number.	As above.
[FunFlux] parameters.<block>: unsupported key(s): <keys>. Supported keys: <list>.	A misspelled key inside a block of parameters.	Use a key from the list.
[FunFlux] parameters.<key>: unknown setting (a typo?). Known settings: ...	A misspelled key directly under parameters.	Use a name from the list.
[FunFlux] parameters.<key>: this setting moved into the block named after your mode: ...	long_read_assembler, flye_input_mode, medaka_model, filtlong or hifiasm at the top of parameters, as in FunFluxL.	Move it into parameters.longread or parameters.hybrid. long_read_assembler is now called assembler.

A misspelled optional key outside parameters is ignored without a message

Only parameters and its blocks are checked for unknown keys. A misspelled optional key in directories, files, links, input or resources is ignored, and the default applies. For example, antismash: instead of antismash_db: in directories means that the antiSMASH database is downloaded, and resources.thread: means 24 threads.

Input files and sample names

These checks look for the input files of the active mode and take the sample names from the file names.

Message	Cause	Fix
[FunFlux] mode=<mode> requires input.<key> to be set in the config.	The key for the input folder of the active mode is missing or empty.	Set input.contigs_dir, input.illumina_dir or input.longread_dir or, in hybrid mode, the last two.
[FunFlux] input.<key>: '<value>' is not an existing directory.	Wrong path. A relative path is read from the launch folder, not from output_dir.	Use an absolute path.
[FunFlux] input.<key>: no input files found. Expected names like <example>.	No file follows the naming pattern.	Name the files as the mode expects: {sample}.fasta (or .fa, .fna) for assemblies, {sample}_R1.fastq.gz and {sample}_R2.fastq.gz for Illumina pairs, and {sample}_ont.fastq.gz or {sample}_pacbio.fastq.gz for long reads.
[FunFlux] input.<key>: more than one file extension in this folder (<a>, <b>). ...	Mixed extensions, or other files in the folder. In contigs mode every file whose name contains a dot is taken as a sample, so a README.txt gives (fasta, txt).	Use one extension for all files of one kind. In hybrid mode the Illumina files and the long-read files may have different ones. Keep other files elsewhere.
[FunFlux] input.<key>: file extension '<ext>' is not recognised (expected one of: ...).	An extension FunFlux does not recognise, for example .fas.	Use fasta, fa or fna for assemblies, and fastq, fq, fastq.gz or fq.gz for reads.
[FunFlux] sample name '<name>' contains characters that are not allowed (...). Rename the input files.	The name holds a space or one of * # @ % ^ / ! ? & : ; | < >. A FASTA in a subfolder of contigs_dir becomes the sample subfolder/name and stops here.	Rename the files. Underscores and dots are allowed. Keep the input folder flat.
[FunFlux] input.illumina_dir: sample '<name>' has an R1 file but no R2 mate: <path>	The R2 file is missing or named differently.	Name the pair {sample}_R1.<ext> and {sample}_R2.<ext>.
[FunFlux] sample '<name>' has both _ont and _pacbio files in <dir>. ...	One sample has files of both long-read technologies.	Keep one long-read file per sample.
[FunFlux] hybrid mode needs Illumina and long reads for every sample.	Some samples have reads of one kind only. The message goes on with No _ont or _pacbio file for:, No Illumina _R1 file for: or both, each followed by the sample names.	Supply both kinds of reads, or run the sample in another batch.

Annotation parameter table

These messages concern the annotation parameter table, its rows and its columns.

Message	Cause	Fix
[FunFlux] files.annotation_params: '<path>' not found.	Wrong path in files.annotation_params.
[FunFlux] files.annotation_params: line N has K tab-separated columns; expected 3 or 4 (sample, species, proteins and, optionally, augustus_species).	Spaces instead of tabs, or a missing or extra column. A species name with a space is still one column.	Separate the columns with tabs.
[FunFlux] files.annotation_params: sample '<name>' names the AUGUSTUS model '<model>', which is not a trained species of the funannotate database at '<db>'. Did you mean '<model>'? Leave column 4 empty, or write '-', to train AUGUSTUS on this genome instead. ...	Column 4 names a model that the database does not hold, for example through a typo, or with capitals where the folder name has none. The message appears at start-up when the database is already on disk. Otherwise it comes from rule augustus_model_check, right after the database is unpacked, and is also written to logs/augustus_model_check_{sample}.log.	Use one of the suggested names, which are the folder names in <db>/trained_species/, or leave column 4 empty to train AUGUSTUS on the genome.
[FunFlux] files.annotation_params: sample '<name>' names the AUGUSTUS model '<model>', which is not a possible model name (letters, digits, '_', '-' and '.' only). ...	Column 4 holds a path, a space or another character that no model name has.	Write the model's folder name only, or leave column 4 empty.
[FunFlux] files.annotation_params: sample '<name>' appears more than once (line N).	A duplicate row.	Keep one row per sample.
[FunFlux] files.annotation_params: sample '<name>' has no row in '<path>'.	Every sample of the batch needs a row. Rows for samples outside the batch are ignored.	Add the row.
[FunFlux] files.annotation_params: protein file '<path>' of sample '<name>' does not exist. (or ... is empty.)	A relative path in the table is read from the launch folder.	Give an absolute path to a non-empty protein FASTA.

A sample whose column 4 is empty can also stop later, during gene prediction, with Not enough gene models <N> to train Augustus; see Other stops in the shared stages.

Databases and links

These messages concern the reference databases and the download links of the configuration.

Message	Cause	Fix
[FunFlux] directories.antismash_db: '<path>' is not a directory. ... (or ... has no 'clusterblast'. ...)	The key must name the antiSMASH databases folder itself, the one that holds clusterblast/, pfam/ and the rest.	Point it there, or leave it empty to download the database.
[FunFlux] directories.unite_db: '<path>' is not a directory. ... (or ... has no 'unite_its_sintax.fasta'. ...)	The folder must hold the UNITE SINTAX reference as an uncompressed unite_its_sintax.fasta.	Unpack or rename the file, or leave the key empty to download it.
[FunFlux] links.unite_its_link: required, unless directories.unite_db names a local UNITE copy.	Both are empty.	Set one of them.
[FunFlux] links.funannotate_link: expected a Zenodo URL of the form https://zenodo.org/records/<id>/files/<archive>, got: <url>	FunFlux finds the checksum file in the same Zenodo record, so the link must be a Zenodo record URL.	Use a Zenodo record URL, as the template does.
[FunFlux] links.funannotate_link: cannot read a file name from the URL: <url>	The URL ends without a file name.	As above.
[FunFlux] links.telomere_db_link: required, but it is missing or empty in the config.	The link to tidk's telomere repeat database is required in every mode. A configuration written before the contig summary existed lacks it.	Copy the line from config/config.yaml.
[FunFlux] parameters.prediction.busco_db: '<name>' is not in the funannotate database at '<db>'. Add it with FunFlux's funannotate conda environment activated ('Other BUSCO sets for gene prediction' in the documentation shows how to find it): funannotate setup -i busco -b <name> -d <db> -w	The funannotate database ships only dikarya.	Run the command shown, in the activated environment (see Other BUSCO sets for gene prediction), or use dikarya.
[FunFlux] parameters.prediction.busco_db: expected a plain dataset name such as dikarya (got: '<value>')	The name holds characters other than letters, digits and underscores.
MissingInputException in rule blob_json ..., in hybrid mode blob_guide_json (from Snakemake)	The screen runs, and nodes.dmp or names.dmp is missing from directories.blast_db. When the folder itself is missing, [FunFlux] NOTE: directories.blast_db '<path>' does not exist on this machine; ... comes first.	Correct the path, or unpack the NCBI taxonomy dump into that folder (see Reference databases).

Screen, masking and BUSCO settings

These messages concern the settings of the screen, of repeat masking, of BUSCO and of the contig summary.

Message	Cause	Fix
[FunFlux] parameters.decontamination.mode: must be one of off | include | exclude (got: <value>)	Another word.	Use one of the three.
[FunFlux] parameters.decontamination.mode: 'auto' is not available in FunFlux. BLAST genus labels are unreliable for poorly sequenced fungi, so auto can remove most of the genome. Use off (the default), read contig_taxonomy_decisions.tsv, then exclude the contaminant genera or include the genera that BLAST assigned to the isolate's own contigs.	mode: auto, which FunFlux 1.x, FunFluxL and FunFluxL+ had, is removed, for the reasons given on Configuration.	Use off. To filter, use exclude or include as the message says.
[FunFlux] parameters.mitochondria.mode: must be one of off | report | separate_confirmed (got: <value>)	Another word, or true or yes, which YAML reads as true. An unquoted off is accepted.	Use one of the three.
[FunFlux] parameters.masking_method: must be tantan | repeatmodeler_repeatmasker (got: <value>)	Another word.	Use one of the two.
[FunFlux] parameters.busco.lineage: must be 'auto' or a BUSCO lineage name such as hypocreales_odb12 (got: <value>). ...	A lineage name must end in _odb and a number. busco --list-datasets prints each name with its version after a dot, such as hypocreales_odb12.2, and the dot and the number after it are not part of the name.	Write auto, or a name such as hypocreales_odb12.
[FunFlux] parameters.contig_summary.telomere_motif: must be auto (the unit is chosen from the data) or the telomere repeat unit of the species, 4 to 30 letters A, C, G and T, such as TTAGGG (got: <value>)	A value that is neither auto nor a plain unit. It may hold a character other than A, C, G and T, for example an IUPAC code such as N, a bracket or a digit. It may also be shorter than 4 or longer than 30 letters.	Write auto, or the unit in plain letters. The known units are listed on Contig summary.
[FunFlux] parameters.contig_summary.telomere_db_sha256: must be the SHA-256 checksum of the telomere database file, 64 hexadecimal characters (got: <value>)	A shortened or mistyped checksum.	Copy the 64 characters from config/config.yaml or, for another version of the database, the ones sha256sum prints for that file (see Telomere repeat database).
[FunFlux] parameters.decontamination.sample_overrides: sample '<sample>' has mode 'auto' in '<path>'. Mode auto is not available in FunFlux. BLAST genus labels are unreliable ... (the rest as in the auto message above)	auto in the mode column of the override table.	Write off, include, exclude or nothing in that cell.
[FunFlux] parameters.decontamination.sample_overrides: sample '<sample>' has mode '<mode>' in '<path>'; the mode column must be off, include, exclude or blank (blank keeps the run-wide mode).	Another word in the mode column. Every row is checked, including the rows of samples outside the batch.
[FunFlux] parameters.decontamination.sample_overrides: sample '<sample>' has discard_no_hit '<value>' in '<path>'; the discard_no_hit column must be true or false (also yes/no, on/off, 1/0) or blank (blank keeps the run-wide setting).	Another word in the discard_no_hit column of a sample of the batch.
[FunFlux] parameters.decontamination.<key>: '<path>' is not an existing file.	include_genera_by_sample, exclude_genera_file or sample_overrides names a file that does not exist. A relative path is read from the launch folder.	Correct the path, or use an absolute one.
[FunFlux] parameters.decontamination.<key>: '<path>' is not UTF-8 text. Save it again as plain text with UTF-8 encoding.	The file has another encoding, for example a table that a spreadsheet program saved as UTF-16.
[FunFlux] parameters.decontamination.<key>: '<path>' must start with a tab-separated header line that has the columns sample and genus (missing: <columns>). (sample and mode for sample_overrides)	The first line lacks a column name, or the columns are separated by spaces.	Start the table with the column names, separated by tabs.
[FunFlux] parameters.decontamination: mode include keeps only the genera you name, but no genus is given for sample(s) <samples>. Name the genus in include_genera, in include_genera_by_sample (one row per sample) or in the include_genera column of sample_overrides.	include is set, run-wide or in sample_overrides, for samples without a genus.

The two decontamination.mode messages can stop any run. The checks of the decontamination files and of include without a genus run only when the screen runs, after the checks of the input folder and of the annotation table.

GeneMark-ES settings

These messages concern GeneMark-ES and its licence key.

Message	Cause	Fix
[FunFlux] parameters.genemark.mode: must be auto | off (got: <value>)	Another word. An empty value (mode: or mode: "") means auto.	Write auto or off.
[FunFlux] directories.genemark_dir: must be an absolute path when GeneMark is enabled.	A relative path.	Use an absolute path.
[FunFlux] directories.genemark_dir: '<dir>' has no executable gmes_petap.pl.	Wrong folder, or the script lacks execute permission.	Point at the folder that holds gmes_petap.pl, or make it executable (chmod u+x).
[FunFlux] GeneMark is enabled but there is no licence key at <home>/.gm_key. ...	GeneMark-ES is on (directories.genemark_dir set, genemark.mode: auto), but the user running Snakemake has no ~/.gm_key.	Install the key (see Licensed and external tools), or set parameters.genemark.mode: off.

Tiberius settings

These messages concern Tiberius, which is on by default, and the download of its model.

Message	Cause	Fix
[FunFlux] parameters.tiberius.mode: must be on | off (got: <value>)	Another word. An empty value means on.	Write on or off.
[FunFlux] parameters.tiberius.model_sha256: expected the SHA-256 checksum of the model archive, 64 characters 0-9 and a-f (got: <value>)	A shortened or mistyped checksum.	Copy the 64 characters sha256sum prints, or leave the key out for the default.
[FunFlux] links.tiberius_model_link: required, but it is missing or empty in the config.	Tiberius is on (the default) and the link is not in the configuration, for example in a configuration written before Tiberius was added.	Copy the line from config/config.yaml, or set parameters.tiberius.mode: off.

SignalP 6 settings

These messages concern SignalP 6 and the DTU archive it is installed from.

Message	Cause	Fix
[FunFlux] parameters.signalp6.mode: must be off | fast | slow | slow-sequential (got: <value>)	Another word.	Use one of the four.
[FunFlux] files.signalp6_archive: required when parameters.signalp6.mode is not off.	SignalP 6 is on without an archive.	Name the DTU archive, or set mode: off.
[FunFlux] files.signalp6_archive: must be an absolute path to the official DTU archive.	A relative path.	Use an absolute path.
[FunFlux] files.signalp6_archive: '<path>' is not a readable file.	Wrong path or permissions.

AMR screen settings

These messages concern the AMR screen (the resistance-mutation screen), which runs ChroQueTas against the FungAMR database.

Message	Cause	Fix
[FunFlux] parameters.amr.chroquetas: must be on | off (got: <value>)	Another word. An empty value means off.	Write on or off.
[FunFlux] parameters.amr.mito_genetic_code: must be auto or the number of an NCBI genetic code: 1-6, 9-16 or 21-33 (got: <value>)	Another word, or a number that is not an NCBI genetic code (7, 8 and 17-20 are none, and miniprot refuses them). An unquoted yes, no, on or off fails too, because YAML reads it as true or false.	Write auto, or the number of the code (see Two runs).
[FunFlux] parameters.amr.chroquetas_script_sha256: must be the SHA-256 checksum of ChroQueTas.sh, 64 hexadecimal characters (got: <value>) (or chroquetas_db_sha256: must be the SHA-256 checksum of the FungAMR database archive, ...)	A shortened or mistyped checksum.	Copy the 64 characters from config/config.yaml, or leave the key out for the default.
[FunFlux] links.chroquetas_link: required, but it is missing or empty in the config.	The AMR screen is on and the link is not in the configuration.	Copy the line from config/config.yaml, or set parameters.amr.chroquetas: off.

Read-based modes

These messages concern the read and assembly settings of the read-based modes. <mode> is longread or hybrid, and each of these blocks is checked only in its own mode. spades_kmers is checked in illumina and hybrid mode.

Message	Cause	Fix
[FunFlux] parameters.spades_kmers: expected auto, or whole numbers separated by commas (got: <value>).	A word other than auto, or a list with another separator.	Write auto, or for example "21,33,55,77,99,127".
[FunFlux] parameters.spades_kmers: every k must be odd and below 128, and the list ascending (got: <value>).	An even k, a k below 1 or above 127, a k given twice, or a list out of order.
[FunFlux] parameters.<mode>.assembler: must be hifiasm | flye (got: <value>)	Another word.
[FunFlux] parameters.<mode>.flye_input_mode: 'pacbio-hifi' is not a value any more. ...	PacBio HiFi reads are recognised from the file name.	Name the reads {sample}_pacbio.fastq.gz and set flye_input_mode: auto.
[FunFlux] parameters.<mode>.flye_input_mode: must be auto | nano-raw | nano-hq (got: <value>)	Another word.
[FunFlux] parameters.<mode>.filtlong.target_bases: must be empty or a size such as 1500000000 or 1.5g (got: <value>)
[FunFlux] parameters.<mode>.hifiasm.genome_size: must be empty or a size such as 40000000 or 40m (got: <value>)
[FunFlux] parameters.<mode>.hifiasm.telomere_motif: must be empty or a DNA/IUPAC motif (got: <value>)
[FunFlux] parameters.<mode>.ont_adapter: write detect, off, or the sequence to trim (20 to 200 bases of A, C, G and T; got: <value>)	Another word, a sequence with other letters, or one shorter than 20 or longer than 200 bases.	Write detect, off or the sequence.
[FunFlux] parameters.<mode>.ont_adapter: <sequence> is a short unit repeated end to end (a telomere repeat, for example), not an adapter. ...	The sequence repeats a unit of 10 bp or less, as a telomere does. Trimming it would cut real genome from the read ends.	Write the adapter sequence, or detect.

Notes that announce a later failure

These NOTE lines do not stop the run. Each one warns of a problem that shows later, in a step or in the result.

Note	What follows
[FunFlux] NOTE: directories.eggnog_db '<path>' does not exist on this machine; ...	eggnog_annotation fails.
[FunFlux] NOTE: files.iprscan '<path>' does not exist on this machine; ...	iprscan_annotation fails.
[FunFlux] NOTE: GeneMark-ES is enabled and needs a valid, unexpired licence key at <home>/.gm_key. ...	An expired key stops the run at genemark_check.
[FunFlux] NOTE: sample name '<name>' is not a valid NCBI locus-tag prefix ...	The run finishes, but the gene IDs cannot be submitted to GenBank as they are. A valid prefix is 3 to 12 letters or digits, starting with a letter.

directories.eggnog_db and files.iprscan are only tested for existence

An eggNOG folder without the data, or an InterProScan installation that does not start, passes these checks and fails when its rule runs.

Failures while the workflow runs

These messages come from single steps after the jobs have started. Each entry gives the message and its cause, and most also say what to do. The step's log holds the tool's own messages.

GeneMark-ES licence

[FunFlux] GeneMark-ES refused to run because of its licence (key file <home>/.gm_key). The key has probably expired: request a new one from the GeneMark site, or set parameters.genemark.mode: off. See <log>.

Cause. Rule genemark_check ran GeneMark-ES on a slice of the masked genome, and the output mentioned the licence. The key in ~/.gm_key has most likely expired (see The licence check before gene prediction).

Fix. Download a new key from the GeneMark site and copy it to ~/.gm_key of the user who runs Snakemake. GeneMark's INSTALL file gives the two commands: gunzip gm_key.gz, then cp gm_key ~/.gm_key. Then run the same Snakemake command again.

To run without GeneMark-ES, set parameters.genemark.mode: off. Gene prediction, and every step after it, then runs again for the samples already predicted with GeneMark-ES.

[FunFlux] GeneMark-ES could not start from <dir>: the Perl named in the first line of gmes_petap.pl (GeneMark ships it as /usr/bin/perl, the computer's own) is missing or lacks a Perl module GeneMark needs. funannotate starts GeneMark the same way, so its gene prediction would fail too. Fix it once, in that folder: perl change_path_in_perl_scripts.pl '/usr/bin/env perl' (GeneMark's scripts then use the Perl of FunFlux's funannotate environment, which has the modules), or install the missing modules into that Perl, or set parameters.genemark.mode: off. See <log>.

Cause. The test output contained Can't locate, which means a missing Perl module, or bad interpreter, which means that the Perl named in the script's first line does not exist. The check starts gmes_petap.pl the way funannotate predict does, so the first line of the script chooses the Perl. GeneMark ships it as #!/usr/bin/perl, which often lacks GeneMark's modules. On a cluster, the compute nodes can have another /usr/bin/perl than the login node.

Fix. Run perl change_path_in_perl_scripts.pl "/usr/bin/env perl" once in the GeneMark folder. The scripts then use the Perl of the funannotate environment, which holds the modules.

[FunFlux] GeneMark-ES could not start from <dir> (a program or a file is missing). See <log>.

Cause. The test output contained command not found or No such file. logs/genemark_check_{sample}.log holds GeneMark's output and names what is missing.

Any other failure on the test slice, for example too little sequence to train on, is logged and the run continues.

SignalP 6

[FunFlux] signalp6_annotation: SignalP 6 is not installed in the conda environment of this rule (the environment was probably rebuilt after SignalP 6 was installed). Delete the file <output_dir>/00.resources/signalp6/installation.json and run the workflow again, so that install_signalp6 installs SignalP 6 into the current environment.

Cause. SignalP 6 is installed into the signalp6 conda environment, and the record of that installation, installation.json, lives in the output folder. A run with another --conda-prefix, from another launch folder, or after .snakemake/conda/ was deleted gets a new environment without SignalP 6. The old record still keeps install_signalp6 from running again.

Fix. Delete 00.resources/signalp6/installation.json and run again. SignalP 6 and funannotate annotate then run again for every sample of the run.

When the installation itself fails, install_signalp6 stops with SignalP 6 installation failed: ... in logs/install_signalp6.log. The rest of the message names the cause:

The message continues with	Cause and fix
the supplied archive does not contain the '<mode>' model (...). Download the DTU archive for the selected mode.	The archive holds the model of another mode. A fast archive such as signalp-6.0i.fast.tar.gz works only with mode: fast. Get the archive for the mode, or change the mode.
configured archive_sha256 does not match the supplied archive (expected ..., observed ...)	files.signalp6_archive names another file than the one the checksum was taken from. Check the path, or update parameters.signalp6.archive_sha256 with what sha256sum <archive> prints.
archive_sha256 must be a 64-character hexadecimal SHA-256 digest	A typo in parameters.signalp6.archive_sha256.
expected exactly one SignalP Python package ..., archive member has an unsafe path: ..., archive contains an unsupported link or special file: ...	The file is not the archive as DTU distributes it. Use the downloaded file unchanged.
the official archive did not install the signalp6 command	The package installed without its command-line program. The pip output is above the message in the same log.

InterProScan

A failed InterProScan job is started once more before the run stops, and logs/iprscan_annotation_{sample}.log keeps the last attempt only.

InterProScan is a Java program, and the memory it may use, its heap, is the -Xmx value on the java line at the end of interproscan.sh in your InterProScan installation. FunFlux does not change it, and resources.ram_gb does not reach it. If logs/iprscan_annotation_{sample}.log shows java.lang.OutOfMemoryError, raise -Xmx there. Keep it below the memory the machine can spare while eggNOG-mapper, antiSMASH and SignalP 6 run beside InterProScan.

After InterProScan ends, the rule checks that iprscan/interproscan_out.xml is complete and is an InterProScan result, and stops otherwise:

InterProScan XML validation failed for <path>: <reason>

Cause. InterProScan exited without an error but left a file that does not parse, or that is not an InterProScan result. Its own messages are above this line in the same log. A full disk is one cause to rule out, as the XML can be a few GB.

Fix. Run again. The rule deletes the old XML before it starts, so a stale file is never taken for a new result.

Could not find a *.proteins.fa file in <dir>/predict_results or update_results. means that funannotate predict wrote no protein file. Its messages are in logs/funannotate_prediction_{sample}.log.

Tiberius

Rule tiberius_model downloads the Tiberius model, and tiberius_prediction runs Tiberius on each sample.

Message	Log	Cause and fix
[FunFlux] tiberius_model: the file downloaded from <link> does not match parameters.tiberius.model_sha256, so it was NOT unpacked ...	tiberius_model.log	The download was cut short, or the authors replaced the model at the same address. Run again for a fresh download. If the checksum still differs, the model has changed, so check it and put its checksum into parameters.tiberius.model_sha256, or set parameters.tiberius.mode: off (see Tiberius model).
curl: (6) Could not resolve host or another curl error	tiberius_model.log	No network access to bioinf.uni-greifswald.de (see Network access).
the job ends with Killed, or Snakemake reports that tiberius_prediction was terminated without a Python error	tiberius_prediction_{sample}.log	The job ran out of memory, as Tiberius needs about 16 GB. Free memory, run fewer jobs at once, or set parameters.tiberius.mode: off.

Telomere repeat database and tidk

Rule telomere_db downloads tidk's telomere repeat database, and telomere_units reads it when it chooses the telomere unit of each sample.

Message	Log	Cause and fix
[FunFlux] telomere_db: the file downloaded from <link> does not match parameters.contig_summary.telomere_db_sha256, so it was NOT used ...	telomere_db.log	The download was cut short, or the link and the checksum do not belong together. The template address names one commit of the database repository, so the file behind it does not change. The download is kept as 00.resources/telomeres/telo_database_v2.tsv.failed. Run again for a fresh download. To accept a file you have looked at, put its checksum, which sha256sum <file> prints, into parameters.contig_summary.telomere_db_sha256. For another version of the database, set the link and its checksum together (see Telomere repeat database).
curl: (6) Could not resolve host or another curl error	telomere_db.log	No network access to raw.githubusercontent.com (see Network access).
[FunFlux] telomere_units: error: ...	telomere_units_{sample}.log	The script stopped on a missing input, a malformed database file or a tidk failure, and tidk's own message follows. A database file that does not parse means that the file at the link has changed. Delete 00.resources/telomeres/telo_database_v2.tsv and run again, and the checksum then says whether it is still the expected file.

UniVec and the adapter screen

Rule univec_illumina downloads NCBI's UniVec for the contig adapter screen; rules blast_adapters and screen_adapters run the screen for every sample.

Message	Log	Cause and fix
[FunFlux] univec_illumina: the file downloaded from <link> does not match parameters.contig_summary.univec_sha256, so it was NOT used ...	univec_illumina.log	The download was cut short, or the link and the checksum do not belong together. The download is kept as 00.resources/adapters/univec_illumina.fasta.download.failed. Run again for a fresh download, or set the link and its checksum together (see UniVec).
curl: (6) Could not resolve host or another curl error	univec_illumina.log	No network access to ftp.ncbi.nlm.nih.gov (see Network access).

ChroQueTas (AMR screen)

With the AMR screen on, rule chroquetas_db downloads ChroQueTas and the FungAMR database, and amr_chroquetas runs for each sample.

Message	Log	Cause and fix
[FunFlux] chroquetas_db: the file downloaded from <link> does not match parameters.amr.chroquetas_script_sha256, so it was NOT used ... (or ... parameters.amr.chroquetas_db_sha256, so it was NOT unpacked ...)	chroquetas_db.log	The download was cut short, or the link and the checksums do not belong together. The template address names one commit, so the files behind it do not change. The download is kept as 00.resources/chroquetas/ChroQueTas.sh.failed or db-v20250811.tgz.failed. Run again for a fresh download. For another version of ChroQueTas, set the link and both checksums together (see ChroQueTas and the FungAMR database).
[FunFlux] chroquetas_db: the database archive from <link> holds no species folder ...	chroquetas_db.log	The link names a commit whose FungAMR_db/db-v20250811.tgz is not the database ChroQueTas expects. Use the template link.
curl: (6) Could not resolve host or another curl error	chroquetas_db.log	No network access to raw.githubusercontent.com (see Network access).
[FunFlux] amr_chroquetas: error: ...	amr_chroquetas_{sample}.log	The script stopped on a missing input file, a database folder without species folders, or a position list of the database (<target>.txt) that does not look like the one of ChroQueTas 1.0.1. It also stops on a path with a space, or when a ChroQueTas run left neither its summary file nor the message that no target could be extracted. The last 20 lines of ChroQueTas's log follow the message. A position list unlike that of ChroQueTas 1.0.1 means another version of the database, so use the template link and checksums. ChroQueTas cannot handle a path with a space, so move the output folder to a path without spaces. ChroQueTas also takes bash, grep -P and other shell commands from the machine, so check them as shown under Commands outside the conda environments.

funannotate database

Rule funannotate_db downloads the funannotate database and unpacks it into directories.funannotate_db, and normalize_funannotate_db repairs the paths stored in it.

Message	Log	Cause and fix
[FunFlux] funannotate_db: <archive> does not match its checksum file <sha256>, so it was NOT unpacked ...	funannotate_db.log	The download was cut short or damaged. Delete both files from directories.funannotate_db and run again. Both are then downloaded afresh.
curl: (6) Could not resolve host or another curl error	funannotate_db.log	No network access to zenodo.org (see Network access). Download the funannotate database by hand on a machine with network access and unpack it into directories.funannotate_db, as described under Downloading it by hand. Then run again. A folder that holds funannotate-db-info.txt is not downloaded again.
ERROR: Funannotate trained-species directory is missing: ..., ERROR: Funannotate database index is missing: ...	normalize_funannotate_db.log	directories.funannotate_db does not hold a complete, unpacked funannotate database. Point it at the folder that holds trained_species/ and funannotate-db-info.txt. If the path is right but the database is incomplete, delete funannotate-db-info.txt and run again, as in the next row.
ERROR: Trained-species asset is missing after extraction: ..., ERROR: Database asset is missing after extraction for archived path: ...	normalize_funannotate_db.log	The unpacked database lacks files. Delete funannotate-db-info.txt from the database folder and run again. The archive kept there is then checked against its checksum and unpacked again, without a new download.
ERROR: [Errno 13] Permission denied: ...	normalize_funannotate_db.log	The database folder is read-only and its stored paths still need repair. Make it writable for one run. Once the paths are right, this step only reads it.

Two runs repairing one funannotate database at once can damage it

Two runs repairing one database at the same time, or one database reached under two different paths, cause damage or repeated work. See Reference databases.

BUSCO downloads

Where BUSCO's lineage data come from depends on parameters.busco.lineage. With auto, every BUSCO job needs network access, while a named lineage needs it once, on the computer that runs Snakemake (see BUSCO on compute nodes without internet).

Setting	Download	Without network access
auto, the default	every BUSCO job downloads the datasets for its sample into 02.assembly/{sample}/eval/busco/busco_tmp/, which is deleted when the job ends, and samples do not share the download. A hypocrealean fungus fetches eukaryota_odb12 (29.1 MB), six placement files (8.3 MB) and hypocreales_odb12 (274.6 MB)	the BUSCO job fails and the run stops, but --keep-going lets the other jobs finish. BUSCO's messages are in logs/completeness_{sample}.log
a lineage, such as hypocreales_odb12	rule busco_data downloads it once per output folder into 00.resources/busco/, on the computer that runs Snakemake, and every BUSCO job reads it offline	busco_data fails, and its messages are in logs/busco_data.log. The BUSCO jobs themselves need no network

With a named lineage, a failed download stops rule busco_data:

Message	Log	Cause and fix
[FunFlux] busco_data: BUSCO did not download the lineage <lineage> (details in <log>). Check parameters.busco.lineage against the list that busco --list-datasets prints, or set it to auto.	busco_data.log	A name BUSCO's server does not know, or no network access to busco-data.ezlab.org.

When BUSCO publishes a newer dataset, a run with auto uses it, and its completeness figures can differ from an earlier run of the same genome. A named lineage stays at the release in 00.resources/busco/ for every later run into the same output folder.

Medaka model

These messages apply to ONT samples in longread and hybrid mode. Rule check_medaka_model checks the Medaka model before the assembler starts. When the check fails, it prints its log, with the table of available models, to the console as well.

ERROR: Medaka could not infer a model from the reads: their FASTQ headers carry no basecaller tag, or one this Medaka version does not recognise. Set parameters.<mode>.medaka_model to one of the consensus models below (pick by your flowcell / device / accuracy), or to FALSE to skip Medaka.

ERROR: Configured medaka_model '<model>' is not available in this Medaka version.

Fix. Use one of these settings:

To	Set
name the model	parameters.<mode>.medaka_model, for example parameters.longread.medaka_model: r1041_e82_400bps_sup_v4.3.0
skip Medaka	parameters.<mode>.medaka_model: false
fall back to the inferred model when a named one is not available	parameters.<mode>.medaka_model_fallback_auto: true

ERROR: Medaka model '<model>' exists, but its weights are not stored on this computer and could not be downloaded (Medaka downloads them from GitHub). ...

Cause. The check downloads the weights of a model Medaka does not ship, so that polishing needs no network, and it runs on the computer that runs Snakemake.

Fix. Start Snakemake on a computer with internet, which on a cluster is the login node. Otherwise name a model whose weights come with Medaka 2.2.2, r1041_e82_400bps_hac_v6.0.0 or r1041_e82_400bps_sup_v5.2.0, or set parameters.<mode>.medaka_model: false.

The Medaka model describes the other messages of the check, what auto needs in the read headers and how to look at them.

Other stops in the shared stages

The steps below run in more than one mode, and each can stop the run. Where a row mentions the log, it is the log of the rule in the Rule column.

Message	Rule	Cause and fix
[FunFlux] filter_contigs: no contig of <sample> is left (kept 0 of M records). ...	filter_contigs	In contigs mode, the FASTA is empty, or its headers are SPAdes-style and no contig reaches 500 bp and 2x coverage. The log says which filter ran. In illumina and hybrid mode, no SPAdes contig reached 500 bp and 2x coverage, so check spades/spades.log and the reads.
[FunFlux] long_read_consensus: Medaka failed for sample <sample> with model '<model>' ...	long_read_consensus	Medaka itself failed, for example for lack of memory or on a damaged read file, after the model had passed check_medaka_model. Medaka's messages are in the same log. parameters.<mode>.medaka_model: false runs without polishing.
the hifiasm rule stops after hifiasm finished, with no [FunFlux] line	long_read_assembly	hifiasm's primary graph holds no contig, so the step that turns it into assembly.fasta stopped the rule. hifiasm's own messages are in the log.
No contigs were kept for sample '<sample>' with mode '<mode>'. ...	select_contigs	The decontamination policy removed every contig, for example include with a genus that BLAST never assigned. Snakemake deletes the outputs of the failed job, the decision table included. contaminants/bestscore.blob.blobDB.table.txt still shows the genus of every contig, and --keep-incomplete keeps the failed job's files. See Decontamination.
Mode 'include' requires at least one genus.	select_contigs	decontamination.mode: include is set without a genus for the sample. The dry run stops on this first, with [FunFlux] parameters.decontamination: mode include keeps only the genera you name, ... (above), so the rule meets it only when a table was changed after the run started.
ERROR: Mitochondrial split would leave an empty nuclear assembly.	separate_mitochondria	Every contig was confirmed mitochondrial, for example in a FASTA that holds only a mitogenome. Run it without the screen, or with mitochondria.mode: report.
[FunFlux] its_windows: ITSx's fungal profile file .../ITSx_db/HMMs/F.hmm is missing or empty. Is the job running in its itsx conda environment (--sdm conda), with the pinned ITSx 1.1.3?	its_windows	The job did not run in the itsx environment, or that environment has no intact ITSx 1.1.3. Run with --sdm conda. If the run already does, delete that environment's folder, which --list-conda-envs shows, so that Snakemake builds it again from the pin file.
[FunFlux] its_windows: no 5.8S profile found in .../ITSx_db/HMMs/F.hmm (no NAME line contains 58S). Does this ITSx version name its 5.8S profiles differently?	its_windows	The ITSx in the itsx environment is not the pinned 1.1.3. Delete that environment's folder, which --list-conda-envs shows, so that Snakemake builds it again from the pin file.
RepeatModeler did not produce a usable repeat library.	repeat_masking	Neither {sample}_rm-families.fa nor consensi.fa.classified exists or holds sequence. RepeatModeler's messages are in the same log. masking_method: tantan does not use RepeatModeler.
Not enough gene models <N> to train Augustus (200 required), exiting (in logs/funannotate_prediction_{sample}.log)	funannotate_prediction	Column 4 of the sample is empty, so funannotate trains AUGUSTUS, and BUSCO found fewer than 200 usable training genes. The assembly may be fragmented, or the fungus far from the BUSCO set (fungi has 290 genes). Name a stored AUGUSTUS model in column 4 (see The AUGUSTUS model), or use a BUSCO set closer to the fungus.
Funannotate did not create annotations.secretome.txt from the supplied SignalP 6 results.	funannotate_annotation	funannotate annotate did not use the SignalP 6 table. Its messages are in the same log.
any MultiQC error	multiqc	Unlike a failed QUAST report, a failed MultiQC report stops the run. See logs/multiqc.log.

Messages that do not stop the run

After these messages the run goes on, so a finished run can still lack a result. The Log column says where each message is written.

Message	Log	Meaning
[FunFlux] genome_assembly_evaluation: QUAST exited with an error for sample <sample>. ...	genome_assembly_evaluation_{sample}.log	QUAST is a report that no other step reads, so the run continues and the MultiQC page lacks the QUAST table for this sample. After fixing the cause, delete 02.assembly/{sample}/eval/quast/ and run again.
[FunFlux] map_evaluation: Qualimap exited with an error for sample <sample>. ... (in hybrid also map_guide_evaluation)	map_evaluation_{sample}.log, map_guide_evaluation_{sample}.log	Qualimap is a report, so the run continues and the MultiQC page lacks the mapping table for this sample. After fixing the cause, delete 02.assembly/{sample}/eval/qualimap/ (or guide/eval/qualimap/) and run again.
[FunFlux] raw_long_read_qc: NanoPlot exited with an error for sample <sample>. ... (or filtered_long_read_qc)	raw_long_read_qc_{sample}.log, filtered_long_read_qc_{sample}.log	NanoPlot is a report, so the run continues and the MultiQC page lacks the read-QC row for this sample. After fixing the cause, delete 01.reads/{sample}/long/raw_qc/ (or filtered_qc/) and run again.
[FunFlux] its_extraction: no 5.8S window was found in the genome of <sample> ...	its_extraction_{sample}.log	No 5.8S gene matched the 358 anchor profiles of ITSx. The ITS files are empty and {sample}_ITS_copies.txt holds only a comment line. A fungus whose 5.8S differs from all of them gives this result.
No ITS1 sequence to classify for sample <sample> ... (or ITS2)	its_classification_{sample}.log	ITSx found none, or every record failed the length guard, which accepts ITS1 and ITS2 of 50-600 bp. {sample}_ITS_copies.txt lists the rejected records.
[FunFlux] telomere_units <sample>: no telomere repeat identified (best candidate <UNIT>: <n> ends in the telomere orientation, <m> opposite)	telomere_units_{sample}.log, and the console	No unit reached 2 contig ends in the telomere orientation with more than twice as many as in the opposite one. telomere_unit.txt reads none, the telomere columns of the contig summary read 0, and chromosomes come out as nuclear_contig. This is usual in a fragmented assembly or in a yeast with irregular telomeres. Set parameters.contig_summary.telomere_motif when the repeat is known (see Telomeres).
[FunFlux] NOTE telomere_units <sample>: ...	telomere_units_{sample}.log, and the console	The chosen telomere unit is unknown to the literature and to the tidk database, or the unit set in parameters.contig_summary.telomere_motif is not the unit the data favour. eval/telomere_candidates.tsv shows every unit with its score.
[FunFlux] NOTE amr_chroquetas <sample>: ...	amr_chroquetas_{sample}.log, and the console	The species in column 2 of the annotation table matches no folder of the FungAMR database, so the sample is not screened. The NOTE names the folders of the same genus, if any. Correct the name when it is misspelt (see The species). 00.resources/chroquetas/species.txt lists the folders. Column 2 is also funannotate's species. A corrected name re-runs augustus_model_check, funannotate_prediction and everything after them, and it goes into the GenBank and GFF files and their names. Change it only when the new name is the right one for the annotation as well.
[FunFlux] NOTE find_ont_adapter <sample>: the reads carry <n> barcodes in at least 10% of the reads with the adapter (...); all of them are trimmed. ...	find_ont_adapter_{sample}.log, and the console	The ONT reads of the sample carry two or more of ONT's published barcodes in quantity, and every one of them was trimmed. If they belong to different samples, demultiplex the reads and run each sample on its own (see Adapter trimming).
[FunFlux] NOTE find_ont_adapter <sample>: the barcode after the adapter is not in ONT's published list; it is trimmed as found in the reads.	find_ont_adapter_{sample}.log, and the console	The reads carry barcode flanks around a barcode ONT does not publish (a custom or new kit). It is trimmed as found; nothing needs fixing unless the barcode should be known.
[FunFlux] NOTE find_ont_adapter <sample>: the reads continue with several different sequences after the adapter that are not published ONT barcodes ...	find_ont_adapter_{sample}.log, and the console	Only the part the sequences share was trimmed, so they stay in the reads, and the adapter screen reports them. Write the whole sequence into parameters.<mode>.ont_adapter to trim it.
[FunFlux] NOTE check_ont_trimming <sample>: more than 1% of the reads kept after trimming still start with the ONT adapter. ...	check_ont_trimming_{sample}.log, and the console	The adapter of these reads may differ from the one FunFlux looks for. The Adapters table of the report and the adapter column of the contig summary show whether adapter reached the assembly.
[FunFlux] screen_adapters <sample>: adapter at <n> contig ends ...	screen_adapters_{sample}.log, and the console	Adapter or barcode sequence lies within 100 bp of these contig ends. 02.assembly/{sample}/eval/adapter_screen.tsv gives the positions. For ONT reads, check that ont_adapter is not off; for an assembly made elsewhere, trim the contig ends before submission. NCBI's FCS-adaptor gives the final check before a GenBank submission.
WARNING: N genera each hold at least 5% of this assembly (...)	select_contigs_{sample}.log	A mixed culture, or one genome split across related genera by BLAST. Read contig_taxonomy_decisions.tsv before trusting the assembly.
WARNING: decontamination is discarding X% of the assembly (...)	select_contigs_{sample}.log	More than 20 % of the bases were removed. If the kept assembly looks incomplete but not contaminated, the contig selector most likely removed genome.
GeneMark-ES stopped on the test slice (exit status N) but not because of its licence or installation; ...	genemark_check_{sample}.log	Gene prediction itself then decides whether GeneMark-ES suits the genome.

On a cluster

Most stops on a cluster come from a value larger than a node or a partition allows, or from a compute node that cannot see a file or the internet. The step's log in logs/ and its Slurm log say which. How to run FunFlux on a cluster is described under On a cluster.

Message	Cause	Fix
maximum number of parallel jobs/used nodes has to be specified for remote execution (use --jobs N with N being a number >= 1)	--executor slurm without --jobs.	Add --jobs N, the number of jobs Snakemake may keep queued at once.
SLURM job submission failed. The error message was sbatch: error: Batch job submission failed: Requested node configuration is not available	A job asks for more CPUs (resources.threads) or more memory (resources.ram_gb) than any node of the partition has.	Lower the value to what sinfo -o "%n %c" shows for CPUs, or sinfo -o "%n %m" for memory in MB (see How the settings reach Slurm).
sbatch: error: Batch job submission failed: Requested time limit is invalid (missing or exceeds some limit), or jobs that stay pending with the reason PartitionTimeLimit in squeue	resources.runtime_hours is longer than the time limit of the partition.	Lower it to what sinfo -o "%P %l" shows, or choose a partition with a longer limit through slurm_partition in --default-resources.
SLURM-job '<id>' failed, SLURM status is: 'OUT_OF_MEMORY'	The step used more memory than it asked for.	The Error in rule <rule>: block names the step. For a heavy step, raise resources.ram_gb. The heavy steps are listed under How the settings reach Slurm. For a light step, raise mem_mb in --default-resources.
SLURM-job '<id>' failed, SLURM status is: 'TIMEOUT'	The step ran past its time limit.	For a heavy or long step, raise resources.runtime_hours. For a light step, raise runtime in --default-resources.
No wall time information given. ... and No job memory information ('mem_mb' or 'mem_mb_per_cpu') is given - submitting without. ... (warnings)	The light steps ask for no time and no memory, so Slurm's defaults apply.	Nothing, if those defaults are a few GB and an hour or more. Otherwise add --default-resources mem_mb=8000 runtime=240.
MissingOutputException ... Missing files after 5 seconds. This might be due to filesystem latency. (or after 60 seconds)	The shared storage shows a file on the login node a while after a compute node wrote it.	Add --latency-wait 60, as in the Run A command. If it is already there, raise it, for example to 120.
a dry run plans finished steps, or download_all -n plans downloads again, with the reason Software environment definition has changed since last execution	The run uses another --conda-prefix than the one the steps ran with.	Use the same --conda-prefix as before, from the same launch folder.
BUSCO (completeness) fails on the compute nodes and logs/completeness_{sample}.log shows a failed download	parameters.busco.lineage: auto needs internet in every BUSCO job.	Name a lineage, and run download_all on the login node (see BUSCO on compute nodes without internet).
[FunFlux] GeneMark is enabled but there is no licence key at <home>/.gm_key in a job's log	The compute node does not see the key in the home folder.	Copy ~/.gm_key to the home folder the compute nodes see (see The GeneMark-ES key).
SSL certificate problem (curl) or CERTIFICATE_VERIFY_FAILED (BUSCO, antiSMASH) in a download log	A network proxy re-signs HTTPS with an in-house certificate, and the conda environments trust only their own list.	Export CURL_CA_BUNDLE and SSL_CERT_FILE before starting Snakemake (see Behind a proxy that re-signs HTTPS).