Skip to contents

Preparing a GenBank submission

Source: vignettes/GenBank-Submission.Rmd
GenBank-Submission.Rmd

RiboPilot writes one folder per export group containing everything you need to submit your records to GenBank: the sequences (with customized FASTA headers), feature tables, and a summary of the checks. This page helps you understand the GenBank submission process. The Export walkthrough shows the Export step itself.

Which files to upload

For rRNA and ITS sequences, NCBI’s Submission Portal (https://submit.ncbi.nlm.nih.gov) takes a FASTA file only. Upload <group>.fasta. The Portal finds the genes itself with ITSx, and writes its own annotation, so the .tbl feature tables are not used there.

The Portal rewrites definition lines in its own wording (small and large subunit ribosomal RNA), so the 18S and 28S wording in your file will not survive.

The FASTA header template

Each record’s FASTA header comes from a template. The default is:

{seqid} [organism={organism}] [moltype=genomic DNA] {organism} {rrna_definition}

and the default for the per-marker files is:

{seqid}_{element} [organism={organism}] {organism} {product}

Edit them in the Export Data window. Each template is checked as you type, and Save template keeps it under a name, so a lab can keep one per submission style. Words in braces are filled in for each record:

Token What it holds
{seqid} The record’s name: the sample ID, or <ID>_p<path>_s<contig> when the sample gives more than one record. Keep it in every template.
{organism} The organism name to submit (see below).
{rrna_definition} The definition phrase, for example “18S ribosomal RNA gene, partial sequence; internal transcribed spacer 1, 5.8S ribosomal RNA gene, and internal transcribed spacer 2, complete sequence; and 28S ribosomal RNA gene, partial sequence.”
{completeness} “complete sequence” or “partial sequence”, for the record as a whole.
{elements} The features in order, for example 18S-ITS1-5.8S-ITS2-28S.
{length}, {n_features}, {missing} Record length, number of features, and missing features.
{sci_name}, {taxid} The name and NCBI taxon ID set in the Taxonomy step.
{element}, {product} Per-marker template only: the marker (18S, ITS1, and so on) and its product name.

Any column of your mapping file works too, as {ColumnName}. The Your mapping columns chips in the Export Data window show every name you can use. You can also use metadata imported from GEOME, GBIF, or NCBI. See Sample metadata.

Required source modifiers

For each rDNA sequences, GenBank requires the organism name, one identifier for the sample (isolate, strain, clone, or specimen voucher), the collection date, and where it was collected. Add them to your template as source modifiers: [specimen_voucher=...], [collection_date=...], and [geo_loc_name=...]. NCBI’s full list is Modifiers for FASTA Definition Lines.

For example, with Voucher, Date, and Country columns in your mapping file:

{seqid} [organism={organism}] [specimen_voucher={Voucher}] [collection_date={Date}] [geo_loc_name={Country}] [moltype=genomic DNA] {organism} {rrna_definition}

Instead of including these values as columns in your mapping file, you can fetch them from GEOME, GBIF, or NCBI BioSample. The Export column of the Metadata button offers ready-made values such as {geome_lat_lon} and {gbif_collection_date}, which you place in the template yourself. Smithsonian NMNH specimens also need a voucherURI; see NMNH voucher submissions.

If a record would get an empty modifier, such as [collection_date=] for a sample with no date, the template check warns and names the records. NCBI does not accept an empty modifier: fill in the column, or use one of NCBI’s accepted missing-value terms, such as missing or not collected. If you continue with export, missing metadata in the FASTA header will be removed on a sample-by-sample basis.

The organism name

{organism} is the name confirmed in the Taxonomy step. If the Taxonomy step did not result in an exact match, {organism} is set to the Taxon label from your map file, so an informal name such as “Upeneus sp. A” is kept. To always use your own label, put {Taxon} in the template instead.

When your map file Taxon differs from the name set in the Taxonomy step, the record’s submission_risk column says taxon_differs(<name>), so you can check which one you want.

Samples with more than one record

When a sample has more than one contig at export, RiboPilot adds [isolate=<ID>_p<path>_s<contig>] to each FASTA header. This allows GenBank to differentiate records that share an organism and a voucher number.

For that reason, a template with its own [isolate=...] is refused for an export group that contains a sample with multiple contigs. It is allowed otherwise.

Complete or partial

Since GenBank performs its own annotation, the following information can usually be ignored for the submission process.

The end of a gene is written as partial (< for the 5’ end or > for the 3’ end in the feature table, and “partial sequence” in the definition line) when:

  • it runs into the end of a contig that has not been trimmed (the gene ran off the assembled contig)
  • the quality check found the gene truncated, meaning less than 90 percent of it was recovered. The end that meets no neighboring spacer or gene is marked if both ends meet one, both are marked
  • a trim cut through the gene. A trim that lands on a gene boundary marks nothing

Your own choice always wins: the 5’ and 3’ buttons under Partial in the contig window (check Edit, pick the feature, then Save) can be used to override RiboPilot. Mark Partial on the contig only changes {completeness} to “partial sequence”; it does not mark any gene end.

Flanking sequence

Contigs often contain sequence beyond the core rDNA region. This is intergenic spacer (IGS): 5’ external transcribed spacer (ETS) before the 18S, and its 3’ ETS after the 28S. RiboPilot does not annotate these flanks, and GenBank tends to reject submissions containing sequences containing IGS/ETS. We recommend that you trim contigs to gene boundaries before export.

Gaps inside a record

A run of 10 or more N bases counts as a gap. NCBI may remove a record with a gap inside the 18S-to-28S region; the gap_in_rDNA flag below marks these.

Check before you submit

After export, the summary files note samples that have submission-risk flags. However, some warning messages may also appear during the export process.

Flagged records. Before it writes anything, the Export window lists every contig that the BLAST screen or the chimera check flagged, and that you have not marked “Reviewed”. You can choose to Leave them out or Export anyway.

The ITSx check. NCBI runs a standard ITSx check on every rRNA and ITS submission. It may suggest that you remove records judged chimeric, gapped, or misassembled by ITSx. RiboPilot runs the same stock ITSx check on every contig when it annotates it, and again when Update on Annotation Review rechecks a contig you trimmed or edited. Export does not run ITSx: it reports the stored result in the itsx_check column of <group>_summary.csv, and the Export done window counts the flagged, clean, and not checked records and names the flagged ones. The ITSx Warnings column in the app shows the same flags. It only warns: nothing is removed.

RiboPilot checks one contig at a time, while NCBI runs ITSx on your whole upload at once. ITSx’s calls can shift a little with the set of sequences it is given, so on borderline records NCBI’s check may differ from RiboPilot’s.

An itsx_check value that starts with not_run means ITSx gave no answer for that record. The record still counts as checked, so it can be locked and exported, and NCBI checks it again when you upload. To check it again in RiboPilot, unlock the contig in Annotation Review, set its State to Ready to run, press Update, and lock it again once its checks are finished.

Value What it means
not_run(ITSx did not finish) ITSx crashed or was stopped before it finished this contig.
not_run(ITSx check failed: <message>) Reading the ITSx result failed; the message says why.
not_run(not checked before upgrade) The contig was locked in an older RiboPilot version that stored no ITSx check for it.
not_run(not checked) No ITSx check is stored for the record.
any other not_run(...) Stored by an older RiboPilot version, for example not_run(no ITSx output).

The submission_risk column. Each record’s row in <group>_summary.csv lists any of these flags:

Flag What it means
chimeric(...) Parts of the sequence match different phyla or kingdoms (a severe chimera call); the reason is in brackets.
chimeric(ITSx) ITSx calls the record chimeric.
gap_in_rDNA A run of N bases falls between the start of the 18S and the end of the 28S.
misassembly Features overlap or are out of order.
off_target(<taxon>) The BLAST screen’s hits agree on a taxon that is not on the sample’s lineage.
contaminant The BLAST screen’s best hit is from a different phylum than the sample.
taxon_differs(<name>) Your Taxon differs from the name set in the Taxonomy step.
missing_organism The record has no organism name.

Check off_target and contaminant records before you submit: the sequence may be another organism’s rDNA, exported under your sample’s name. Weaker chimera evidence is in the separate chimera_warning column; it is not a submission risk, but it is worth a look in the app.

What is in the export folder 

out/export/<group>/
    <group>.fasta           every record, one file (this is what you upload)
    <group>.tbl             feature tables (for your records; not uploaded)
    <group>_summary.csv     one row per record, with submission_risk
    <group>_report.html     review report with feature maps and alignments
    README.txt              how to submit
    samples/                one .fasta and .tbl per record
    markers/                one folder per marker (18S, ITS1, and so on)

The summary CSV also says which records are variants of each other, in the variant_class, variant_group, variant_identity, and n_variants columns; see Quality checks and flags.