CLI: lookup commands¶
Six commands that answer from a published table rather than from one genome's files. tf
names an assembly, xref and homology name a species, and motif names a FASTA. Every
answer says which publisher and which release produced it.
$ genome xref ids "Homo sapiens" --from-stems symbol ENSG00000141510
gene id stems -> symbol ids for Homo sapiens (hgnc 2026-07-07)
source https://storage.googleapis.com/public-download-files/hgnc/archive/archive/quarterly/tsv/hgnc_complete_set_2026-07-07.txt
1 resolved, 1 symbol ids, 0 this release names none for
ENSG00000141510 TP53
The tables land in the shared data directory, where every project on the machine reads the same copy.
Each synopsis below is its group's own --help, so it leaves the leading genome off:
xref ids is run as genome xref ids. homology and motif carry one command each, which
their synopses fold into the group name; they are run as genome homology links and genome
motif scan. --json, exit codes and the split between stdout and stderr are on the
CLI overview.
genome tf¶
gene-list prints the gene ids a published census judges transcription factors, and
cofactor-list the ids a publisher lists as cofactors. Which census or table answers is
decided by the species the assembly's own metadata row names, never by anything you pass.
Both resolve the publisher's gene id stems into the ids the annotation actually spells, so
the output joins straight to a counts matrix. Who published the verdict, and what it does
and does not cover, is on Transcription factors.
$ genome tf cofactor-list ce11 > cofactors.txt
TF cofactors for ce11 / wormbase_ws298 (Caenorhabditis elegans)
AnimalTFDB 4.0 (PMID 36268869) — https://guolab.wchscu.cn/AnimalTFDB4_static/download/Cof_list_final/Caenorhabditis_elegans_Cof
317 cofactors, 317 gene ids, 0 stems this annotation carries no gene for
A worm assembly is answered by cofactor-list and refused by gene-list, because
AnimalTFDB assessed worm cofactors and no publisher has released a worm TF census. The
message names the species that do have one. An assembly that is not registered here, and one
whose species nothing names, exit 1 too, each with its own message.
The two commands, with every argument and option:
tf¶
Read a species' transcription factors and cofactors in one annotation's gene ids.
Usage¶
tf [OPTIONS] COMMAND [ARGS]...
Arguments¶
No arguments available
Options¶
| Name | Description | Required | Default |
|---|---|---|---|
--install-completion |
Install completion for the current shell. | No | - |
--show-completion |
Show completion for the current shell, to copy it or customize the installation. | No | - |
Commands¶
| Name | Description |
|---|---|
gene-list |
Print the gene ids a published census... |
cofactor-list |
Print the gene ids a publisher lists as... |
Subcommands¶
gene-list¶
Print the gene ids a published census judges transcription factors, one per line.
Nothing here decides what a transcription factor is. The verdict is the census's — Lambert et al. 2018 for human, AnimalTFDB 4.0 for mouse — and which one spoke is printed beside the answer, since citing it is the condition on shipping it. The species comes from the assembly's own metadata row and is never passed in, so asking for human transcription factors while holding a mouse assembly is not expressible.
A census is keyed by gene id stems — gene ids with the version suffix dropped — and a registered annotation is not, so every stem is resolved into the ids that annotation actually spells and the output joins to a counts matrix with nothing left to normalise. A stem naming two genes prints both rather than one of them.
Only the ids go to stdout, so the output pipes: the heading, the census's attribution
and the counts go to stderr, the last of them saying how many stems this annotation
carries no gene for. --json carries the whole record — every gene with the census's
own assessment and DBD family, the provenance to cite, and those unresolved stems.
Assessed-positive genes only, and there is no flag to widen it: a gene the census
assessed and turned down is a verdict too, but a bare id list has nowhere to say which
of the two an id is, and a pipeline would read the rejected ones as transcription
factors. Genome( is where that answer
is expressible, because there each id travels with the verdict reached on it.
Exits with code 1 when the annotation is not registered here, when no census ships for the assembly's species, and when nothing says what species the assembly is — three different facts, each with its own message, and none of them an empty list of genes.
Usage¶
tf gene-list [OPTIONS] ASSEMBLY
Arguments¶
| Name | Description | Required |
|---|---|---|
ASSEMBLY |
Assembly name, e.g. 'hg38'. | Yes |
Options¶
| Name | Description | Required | Default |
|---|---|---|---|
--annotation TEXT |
Ask about this registered annotation instead of the assembly's default one. An assembly with no default and none named has nothing to answer about, and says so. | No | - |
--json |
Emit JSON instead of plain text. | No | False |
cofactor-list¶
Print the gene ids a publisher lists as transcription cofactors, one per line.
genome tf gene-list for the other half of the machinery, and the same shape: a
cofactor — a chromatin remodeller, a histone-modifying enzyme, a Mediator subunit —
recognises no sequence of its own and so has no motif to scan for, but which genes are
cofactors is published, and this is that list met with one annotation. Nothing here
decides what a cofactor is: membership and classification both travel with the
publisher, and who to cite is printed beside the answer.
The species comes from the assembly's own metadata row and is never passed in, so asking for mouse cofactors while holding a worm assembly is not expressible. A table is keyed by gene id stems — gene ids with the version suffix dropped — and a registered annotation is not, so every stem is resolved into the ids that annotation actually spells and the output joins to a counts matrix with nothing left to normalise. A stem naming two genes prints both rather than one of them.
Only the ids go to stdout, so the output pipes: the heading, the publishers'
attribution and the counts go to stderr, the last of them saying how many stems this
annotation carries no gene for. --json carries the whole record — every gene with
the publisher that listed it and that publisher's own classification, one provenance
entry per publisher to cite, and those unresolved stems.
A worm assembly is answered here and refused by genome tf gene-list: a publisher
assessed worm cofactors and none has released a worm TF census. That is what the
publishers have done rather than a defect here.
Exits with code 1 when the annotation is not registered here, when no cofactor table ships for the assembly's species, and when nothing says what species the assembly is — three different facts, each with its own message, and none of them an empty list of genes.
Usage¶
tf cofactor-list [OPTIONS] ASSEMBLY
Arguments¶
| Name | Description | Required |
|---|---|---|
ASSEMBLY |
Assembly name, e.g. 'mm39'. | Yes |
Options¶
| Name | Description | Required | Default |
|---|---|---|---|
--annotation TEXT |
Ask about this registered annotation instead of the assembly's default one. An assembly with no default and none named has nothing to answer about, and says so. | No | - |
--json |
Emit JSON instead of plain text. | No | False |
genome xref¶
ids converts identifiers to and from gene id stems, which is how a column of Entrez
GeneIDs from a GEO series or UniProt accessions from a mass-spec run reaches the rest of
this package. The direction is named and never inferred: --to-stems NAMESPACE reads the
ids as that namespace, --from-stems NAMESPACE answers in it, and naming neither or both
exits 2. What the namespaces are and which source answers by default is on
Gene identifiers.
$ genome xref ids "Homo sapiens" --to-stems hgnc HGNC:11998 HGNC:13666 HGNC:10041 > stems.tsv
hgnc ids -> gene id stems for Homo sapiens (alliance 9.0.0)
source https://download.alliancegenome.org/9.0.0/GENECROSSREFERENCE/COMBINED/GENECROSSREFERENCE_COMBINED_11.tsv.gz
2 resolved, 3 gene id stems, 1 this release names none for
$ cat stems.tsv
HGNC:11998 ENSG00000141510
HGNC:13666 ENSG00000094914
HGNC:13666 ENSG00000291836
HGNC:10041
Every id you passed gets at least one row. One naming two genes prints both rather than whichever came first, and one this release names nothing for gets a row with an empty second column, so nothing you asked about goes missing from the output.
symbols is the separate command for going the other way from a gene symbol. It matches
approved, previous and alias spellings, and each row says which kind matched, which is why
it is not a third direction of ids:
$ genome xref symbols "Homo sapiens" ARNTL ADCY3 Brca1 > genes.tsv
gene symbols -> gene id stems for Homo sapiens (hgnc 2026-07-07)
source https://storage.googleapis.com/public-download-files/hgnc/archive/archive/quarterly/tsv/hgnc_complete_set_2026-07-07.txt
columns asked, symbol, gene_id_stem, kind
matching exact, on approved, previous, alias spellings
2 resolved, 3 matches, 1 this release matched nothing for
$ cat genes.tsv
ARNTL ARNTL ENSG00000133794 previous
ADCY3 ADCY3 ENSG00000138031 approved
ADCY3 ADCY3 ENSG00000155897 previous
Brca1
ARNTL is a spelling HGNC retired and it still reaches its gene. Matching is exact by
default, because the species already fixes the authority; --case-insensitive folds both
sides and still answers with every gene it matched.
The two commands, with every argument and option:
xref¶
Convert identifiers against one published xref set, and match gene symbols.
Usage¶
xref [OPTIONS] COMMAND [ARGS]...
Arguments¶
No arguments available
Options¶
| Name | Description | Required | Default |
|---|---|---|---|
--install-completion |
Install completion for the current shell. | No | - |
--show-completion |
Show completion for the current shell, to copy it or customize the installation. | No | - |
Commands¶
| Name | Description |
|---|---|
ids |
Convert identifiers to and from gene id... |
symbols |
Print the genes each gene symbol names,... |
Subcommands¶
ids¶
Convert identifiers to and from gene id stems against one published xref set.
The way a column of Entrez GeneIDs from a GEO series, UniProt accessions from a mass-spec run or HGNC ids from a curated resource reaches this package's answers, without writing Python and without the hand-built join everyone in the lab writes slightly differently. No assembly is named and no genome is opened: an identifier is a name and not a place.
The direction is named, never inferred. --to-stems NAMESPACE reads the ids as
that namespace and answers in gene id stems; --from-stems NAMESPACE reads them as
stems and answers in that namespace. A string does not say which system it belongs to,
so HGNC:11998 asked the wrong way answers nothing found rather than quietly turning
around. There is no third direction: Entrez to HGNC is two calls and the join is yours,
which keeps the hop visible in your pipeline rather than invisible in ours.
A gene symbol is the one namespace these two directions do not mirror.
--from-stems symbol is answered here and gives the authority's single current approved
spelling. The other way round is genome xref symbols, because a symbol also matches
spellings the authority has retired and each match carries which kind it was — so
--to-stems symbol exits 2 naming that command rather than matching approved spellings
alone, which is what drops 31 of EpiFactors' 801 rows.
The pairs go to stdout, tab-separated, so the output pipes — cut -f2 is the
answer, cut -f1 says what asked for it — and the heading, the publisher's URL and the
counts go to stderr. An id naming two genes prints two rows rather than whichever came
first, and an id that resolved to nothing gets a row too, with an empty second
column: what your list holds and this release does not is the one thing a hand-rolled
join drops silently. --json carries the same answer, keyed by what was asked about,
with those ids under unresolved.
Omitting --source answers from the species' default xref source, so everyone in the
lab reaches for the same one without discussing it. It is a default and not a
recommendation: naming a source is how the scientific choice gets made deliberately,
and every answer names the source and the release that produced it either way. A
default is per species and per question, so --from-stems symbol — the one question
here that is about symbols — is answered by the source that carries them, hgnc for
human and alliance_bgi for mouse and worm, rather than by the identifier default.
Naming a species prepares its set, which the first time is a download. The lab's CPU cluster compute nodes have no internet, so a set must be constructed once from a login node — by running this there, or from Python — before a job that needs it is submitted; after that it is read from the Data dir and shared by every project on the machine.
Exits with code 2 when no direction is named or both are, and when the symbol namespace is asked toward the hub; and with code 1 when no set exists for the species — the message names the ones that do — when the source is not one this package prepares, when the set is not here and cannot be fetched, when the namespace is not one the set carries, and when a directory holds a set left unfinished.
Usage¶
xref ids [OPTIONS] SPECIES IDS...
Arguments¶
| Name | Description | Required |
|---|---|---|
SPECIES |
Species an xref set exists for, e.g. 'Homo sapiens' — the slug 'homo_sapiens' names the same one. A species none exists for names the ones that do rather than answering nothing. | Yes |
IDS |
The identifiers to convert. Each comes back in the order you passed it, with its version suffix and its namespace's CURIE prefix accepted either way. | Yes |
Options¶
| Name | Description | Required | Default |
|---|---|---|---|
--to-stems NAMESPACE |
Read the ids as this namespace and answer in gene id stems: ensembl, entrez, uniprot, hgnc, mgi, wormbase, whichever of them this set carries. A gene symbol is not among them — it matches spellings the authority has retired and each match carries which kind it was, so genome xref symbols answers it. Exactly one of this and --from-stems is named. |
No | - |
--from-stems NAMESPACE |
Read the ids as gene id stems and answer in this namespace, symbol included — which gives the authority's one current approved spelling, the one a figure axis wants. A versioned gene id is accepted and reduced to its stem, so an annotation's own ids go straight in. |
No | - |
--source TEXT |
Answer from this xref source rather than the species' default one. Which publisher answers is a scientific choice and not a detail: NCBI and Ensembl agree on 57.6% of human gene-level (GeneID, ENSG) pairs. | No | - |
--json |
Emit JSON instead of plain text. | No | False |
symbols¶
Print the genes each gene symbol names, and which kind of spelling matched.
The way a gene list copied out of a paper becomes usable without first finding its ids — and without the join that silently drops every row spelling its gene the way the authority used to. No assembly is named and no genome is opened: a symbol is a name and not a place.
A symbol is matched, never converted. Approved, previous and alias spellings are
matched, every Gene id stem any of them names comes back, and each match says which
kind of spelling it was — so ambiguity is what you are handed rather than something
resolved on your behalf. ADCY3 is HGNC's approved symbol for one gene and a symbol it
retired from another, and both are printed. This is why it is a command of its own and
not a third direction of genome xref ids: matching approved spellings alone would drop
exactly the rows this exists for — 31 of EpiFactors' 801 human rows spell their gene the
way HGNC spelled it years ago. The opposite hop, a stem to the authority's one current
approved spelling, is genome xref ids --from-stems symbol.
The matches go to stdout, tab-separated, so the output pipes — cut -f3 is the
answer, cut -f1 says what asked for it and cut -f4 says which kind of spelling
matched — and the heading, the publisher's URL, the counts and what this source could
not have matched go to stderr. A symbol naming two genes prints two rows rather than
whichever came first, and a symbol this release matched nothing for gets a row too,
with every other column empty. Column 2 is the authority's own spelling, which is not
always the one asked about: folded, brca1 asked comes back as BRCA1 matched.
Matching is exact by default, because the species is fixed by the set:
--case-insensitive folds both sides and still answers with every gene matched rather
than picking one.
What this source could not have matched is printed too. Only HGNC publishes previous and alias spellings typed; mouse and worm match current approved symbols alone, their authorities' typed spellings belonging to publishers that cannot be pinned or cannot be fetched. So the answer says which kinds it could match and why the others are missing — without which this gene is not in the release and this source does not publish the spelling you used would both be silence.
Omitting --source answers from the species' default source for symbols, which is
not the same row as its default for identifiers: human's identifiers come from
alliance, whose cross-reference file publishes no human symbol at all, and its symbols
from hgnc; mouse's and worm's from alliance_bgi. A default is per species and per
question for that reason, and every answer names the source and release that produced it
either way. Naming a source is still how the scientific choice gets made deliberately,
and a named one is never swapped — so --source alliance here exits 1 saying that set
carries no symbol, rather than quietly answering from somebody else's file.
Naming a species prepares its set, which the first time is a download. The lab's CPU cluster compute nodes have no internet, so a set must be constructed once from a login node — by running this there, or from Python — before a job that needs it is submitted.
Exits with code 1 when no set exists for the species — the message names the ones that do — when the source is not one this package prepares, when a named source carries no symbols at all — the message names the one that does — when the set is not here and cannot be fetched, and when a directory holds a set left unfinished.
Usage¶
xref symbols [OPTIONS] SPECIES SYMBOLS...
Arguments¶
| Name | Description | Required |
|---|---|---|
SPECIES |
Species an xref set exists for, e.g. 'Homo sapiens' — the slug 'homo_sapiens' names the same one. The species fixes the authority, so a symbol is matched against that authority's spellings and no other's. | Yes |
SYMBOLS |
The gene symbols to match, answered in the order you passed them. Surrounding whitespace goes; case does not, unless --case-insensitive is named. | Yes |
Options¶
| Name | Description | Required | Default |
|---|---|---|---|
--source TEXT |
Answer from this xref source rather than the species' default one for symbols, which is hgnc for human and alliance_bgi for mouse and worm. Naming one is deliberate and is never overridden: a source that carries no symbol says so and names the one that does, rather than matching nothing. |
No | - |
--case-insensitive |
Fold case on both sides — your spelling and the authority's — and still answer with every gene matched. Off by default: the species is fixed by the set, so 'Brca1' asked of a human set is a mouse spelling asked of the wrong authority. | No | False |
--json |
Emit JSON instead of plain text. | No | False |
genome homology¶
links prints the genes of another species that a gene id stem's gene is homologous to, on
Ensembl Compara's own gene trees. Any pairing among human, mouse and worm answers, either
way round, off one file fetched once and read locally. What the labels mean and how to put
an answer back into an annotation's own gene ids is on Homology.
$ genome homology links "Caenorhabditis elegans" "Homo sapiens" \
WBGene00020462 WBGene00008317 WBGene00008352 > homologs.tsv
Caenorhabditis elegans -> Homo sapiens orthologs (Ensembl Compara 116)
source Ensembl Compara release 116 (PMID 26896847) — https://ftp.ensembl.org/pub/release-116/tsv/ensembl-compara/homologies/homo_sapiens/Compara.116.protein_default.homologies.tsv.gz
columns gene_id_stem, homolog_gene_id_stem, homology_type, is_ortholog, is_high_confidence, goc_score, wga_coverage
2 resolved, 2 links, 1 this release names no homolog for, 0 dropped partners
quality goc_score and wga_coverage null on every link of this set, so a filter on either empties rather than narrowing
$ cat homologs.tsv
WBGene00020462 ENSG00000177479 ortholog_one2one True True NULL NULL
WBGene00008317 ENSG00000164074 ortholog_one2one True True NULL NULL
WBGene00008352
Every cell is the publisher's, printed verbatim and never recomputed. An empty row is not
NULL: an empty row is a gene this release names no homolog for, and NULL is Compara's
own word for a cell it recorded nothing in on a link that does exist. Orthologs are the
answer by default and --paralogs returns every link the publisher wrote, marked by its own
homology_type rather than filtered out.
The command, with every argument and option:
homology¶
Print the genes of another species a gene id stem's gene is homologous to.
The way a hit carries across species without leaving the package and without the Ensembl BioMart web API, whose intermittent failures make a pipeline built on it fail irreproducibly. Everything here is a bulk file fetched once and read locally. No assembly is named and no genome is opened: a Homology set is anchored to a species pair and a release, not to a build.
Every cell is Ensembl Compara's. The homology_type — ortholog_one2one,
ortholog_one2many, ortholog_many2many — is the publisher's own tree-derived label
printed verbatim, and it is never recomputed from what came back: an answer can show
one partner and still read ortholog_one2many, which is the point of carrying the
label rather than counting rows. The high-confidence flag and both quality scores come
through the same way, and this package publishes no score, ranking or "best ortholog"
of its own.
The links go to stdout, tab-separated, so the output pipes — cut -f2 is the
answer, cut -f1 says what asked for it — and the heading, the attribution, the counts
and the two qualifications below go to stderr. A gene with three homologs prints three
rows rather than whichever came first, and a stem this release names no homolog for
gets a row too, with every other column empty: what your list holds and this release
does not is visible rather than dropped. An empty cell there is not NULL, which is
the publisher's own word for a cell it recorded nothing in on a link that does exist.
Two qualifications ride on every answer. The Dropped partners — the homologous
genes a filter removed — are counted and named, so a link that merely looks
one-to-one in your view stays distinguishable from one the publisher called one-to-one.
And whichever quality columns this set holds no value in anywhere are named up front:
Compara records neither goc_score nor wga_coverage on any link of either worm
pairing, so a filter written against one would empty itself in silence.
Orthologs are the answer by default and --paralogs returns every link the publisher
wrote; a Paralogy link is marked by its own homology_type rather than excluded,
so not an ortholog stays distinguishable from absent. Release 116 publishes no
cross-species paralogy for these three species, so on it the flag changes nothing.
Naming a pair prepares its set, which the first time is a download. The lab's CPU cluster compute nodes have no internet, so a set must be constructed once from a login node — by running this there, or from Python — before a job that needs it is submitted; after that it is read from the Data dir and shared by every project on the machine.
Exits with code 1 when no set is pinned for the species — the message names the ones that are — when the release is not pinned, when both species are the same one, when a stem carries a version, when the set is not here and cannot be fetched, when a directory holds a set left unfinished, and when the file that was recorded as holding this pair holds none of its rows, which means Compara re-partitioned and the message names the other file.
Usage¶
homology [OPTIONS] SPECIES OTHER_SPECIES STEMS...
Arguments¶
| Name | Description | Required |
|---|---|---|
SPECIES |
Species the gene id stems belong to, e.g. 'Homo sapiens' — the slug 'homo_sapiens' names the same one. A species no set is pinned for names the ones that are rather than answering nothing. | Yes |
OTHER_SPECIES |
Species the homologous genes come back in. Any pairing among human, mouse and worm; the same species twice is refused, since a gene's paralogs within one species is a different question this does not answer. | Yes |
STEMS |
The gene id stems to ask about, answered in the order you passed them. Compara writes its gene ids bare, so a versioned id is refused by name rather than answered emptily. | Yes |
Options¶
| Name | Description | Required | Default |
|---|---|---|---|
--release TEXT |
Ensembl Compara release to answer from. Recorded on the answer, so a result is reproducible a year later. | No | 116 |
--paralogs |
Return every link the publisher wrote for these genes rather than only the ones its own label calls a speciation event. A paralogy link is marked by that label in the homology_type column, never excluded. |
No | False |
--json |
Emit JSON instead of plain text. | No | False |
--install-completion |
Install completion for the current shell. | No | - |
--show-completion |
Show completion for the current shell, to copy it or customize the installation. | No | - |
Commands¶
No commands available
genome motif¶
scan reads a FASTA, scores every sequence against a JASPAR release, writes the hits to
Parquet and prints a summary of the run. It is the batch case and the only motif command.
Listing, plotting and comparing motifs are notebook work, on
Motifs.
$ genome motif scan peaks.fa hits.parquet --release 2024
scanned 500 sequences with 781 motifs from JASPAR 2024 vertebrates
background 0.298, 0.198, 0.198, 0.306
threshold 0.0001
skipped 98 under 7 positions, so not scanned: MA0004.1, MA0130.1, MA0151.1, …
workers 14
hits 22251 -> hits.parquet
The skipped list is trimmed here; the command names all 98. A motif under seven positions cannot reach the default threshold at all, so it is named rather than scanned at some looser cutoff you did not ask for.
The hits go to the named file and the summary to stdout, so --json is never corrupted by
table data. Read the file back with genome.tf.motif.read_hits, which restores the compact
dtypes and the provenance that says what the scan was; pandas.read_parquet gives the rows
and drops both. --background decides the answer more than any other option, and
whichever mode was used is recorded in the summary and on the hits.
The command, with every argument and option:
motif¶
Scan a FASTA with a JASPAR release and write the hits to Parquet.
The batch case, and the one motif operation that belongs in a shell script and a scheduler job: a FASTA in, a Parquet file out, a summary of the run on standard output. Listing, plotting and comparing motifs are notebook work and get no command.
The hits go to the named file and the summary to standard output, so --json is
never corrupted by table data. The summary says which release was scanned with, how
many motifs it scanned and which it left out, the background actually used, how many
sequences were read, and how many hits were written where — the same facts the table
itself carries, so a pipeline consuming the summary and a reader opening the file
months later agree about what happened.
It defaults to every core the allocation granted, where the library defaults to
one: a console script is a proper entry point, so the process-pool hazard that
justifies the serial default does not apply here. --workers 1 scans serially and
answers with the identical table.
Naming a release prepares it, which the first time is a download. The lab's CPU cluster compute nodes have no internet, so a release must be constructed once from a login node — by running this there, or from Python — before a job that needs it is submitted; after that it is read from the Data dir and shared by every project on the machine.
Exits with code 1 when the FASTA is not there or is not FASTA, when the release or tax
group is not one this package prepares, when the threshold is not a p-value in (0, 1),
when the worker count is below 1, when the release is not prepared here and cannot be
fetched — which is what a compute node with no internet looks like — and when a directory
holds a release left unfinished.
Usage¶
motif [OPTIONS] FASTA OUTPUT
Arguments¶
| Name | Description | Required |
|---|---|---|
FASTA |
FASTA to scan, plain or .gz. A record is named by its header up to the first whitespace — what the aligners write into an alignment made from the same file. | Yes |
OUTPUT |
Where to write the hits, as Parquet. Read it back with genome.tf.motif.read_hits, which restores the dtypes and the provenance both. | Yes |
Options¶
| Name | Description | Required | Default |
|---|---|---|---|
--release TEXT |
JASPAR release to scan with: 2024, 2026. Recorded on the hits, so a table opened months later still says what made it. | No | 2026 |
--tax-group TEXT |
JASPAR taxonomic group: vertebrates, plants, insects, nematodes, fungi, urochordates, diatoms, all. It chooses which file is fetched rather than filtering one afterwards, so a worm scan never pays for a thousand plant matrices. | No | vertebrates |
--threshold FLOAT |
The per-position p-value each motif's cutoff is converted from — one number meaning the same stringency for a short matrix and a long one. A motif that cannot reach it is left out and named among the skipped, never called at something looser. | No | 0.0001 |
--background [auto|uniform|derive] |
The base composition scores are taken against, and the parameter that decides the answer most: auto derives it from the input above 10 000 unambiguous bases and stays uniform below that, uniform pins it, derive derives whatever the input holds. Four frequencies of your own are a Python call rather than a flag. | No | auto |
--workers INTEGER |
How many processes to shard the scan across. Every core the allocation granted by default — the Slurm allocation first, then this process's CPU affinity, then the machine — where the library defaults to one. More than one produces the identical table. | No | - |
--json |
Emit JSON instead of plain text. | No | False |
--install-completion |
Install completion for the current shell. | No | - |
--show-completion |
Show completion for the current shell, to copy it or customize the installation. | No | - |
Commands¶
No commands available