FASTA vs. GenBank vs. TSV: Converting Between Sequence File Formats

What each format actually stores

FASTA is the simplest and most universal sequence format: a single header line starting with ">" followed by the raw sequence, wrapped or unwrapped. It carries no annotation at all — no gene boundaries, no strand information, nothing but the letters themselves. That's exactly why most alignment, search and primer-design tools accept FASTA as their baseline input: they only need the sequence, and FASTA guarantees there's nothing else to parse.

GenBank (.gb/.gbk) is the annotated flat-file format. Beyond the raw sequence in its ORIGIN block, it carries a FEATURES table listing genes, CDS regions, promoters, primer-binding sites and other elements, each with explicit coordinates (for example CDS 1..723, or CDS complement(100..822) if the feature reads off the reverse strand). You reach for GenBank specifically when the annotation is the point — tracking where a CDS starts, where a tag ends, where a promoter sits relative to your insert.

Tab-separated (TSV) output is different again: one row per sequence record, with columns like id, length, GC% or other computed values. Neither FASTA's header-plus-sequence block nor GenBank's LOCUS/FEATURES/ORIGIN structure is tabular, so neither drops cleanly into a spreadsheet or a script that expects rows and columns — TSV exists to bridge that gap.

Converting FASTA to GenBank

Because FASTA never had annotation to begin with, converting it to GenBank can't invent any. What you get by default is an unannotated GenBank shell — a LOCUS line and an ORIGIN block with your sequence, and little or nothing in the FEATURES table. If you need real annotation in the output (a CDS at specific coordinates, a source organism, a gene name), you have to supply that yourself as part of the conversion; the tool can only arrange what you give it, not guess where your coding region starts.

Converting GenBank to FASTA

This direction is simpler because you're removing information, not adding it. Converting GenBank to FASTA strips the FEATURES table and keeps only the raw sequence, typically using the LOCUS or DEFINITION line to build the FASTA header. That's the right move whenever the downstream tool only reads FASTA and doesn't care about feature coordinates — BLAST-style searches, most aligners, and many primer or oligo tools all fall into this category. The tradeoff is permanent for that file: once you've dropped the FEATURES table, you'd need to go back to the original GenBank record to recover it.

Extracting a CDS or protein sequence from a GenBank record

This is the practical route from "a full annotated gene or plasmid record" to "just the coding sequence or its protein," without manually counting bases against the FEATURES table by hand — a step that's easy to get off by one base, especially across a complement-strand feature.

1. Find the CDS feature you want in the FEATURES table and note its coordinates, e.g. CDS 145..912 or CDS complement(145..912).
2. Slice exactly that coordinate range out of the ORIGIN sequence — this gives you the coding sequence on its own, separate from the rest of the plasmid or contig.
3. If the feature is marked complement(), reverse-complement the sequence you just sliced out — the coordinates describe positions on the sequence as written in ORIGIN, and complement() means the gene actually reads on the opposite strand.
4. Optionally translate that CDS using the standard codon table if what you actually need is the protein sequence, not the nucleotides.

Choosing a format for the task in front of you

• Keep working sequences in GenBank whenever annotation matters — cloning design, feature-aware analysis, anything where you need to know where the CDS or promoter actually sits.
• Convert down to FASTA when a tool only needs the raw sequence and has no use for feature coordinates.
• Extract to TSV when you need many records' worth of one or two computed values — length, GC%, accession — in a table a spreadsheet or script can consume directly.

Converting without doing it by hand

SeqBench's Format Converter handles these conversions directly: FASTA, GenBank and TSV in either direction, plus pulling out just the CDS or translated protein sequence from a GenBank record's FEATURES table, complement strand included. If you're starting from an accession rather than a local file, Sequence Fetcher retrieves the GenBank or FASTA record for a given accession before you convert or extract from it. And if you want to see the feature coordinates laid out on a map before you extract anything, GenBank Viewer renders the annotated record as a linear or circular map with its feature table alongside.

Related references

Related tools

Related guides

• How to Read a Plasmid Map (and the GenBank Format Behind It) →
• How to Fetch a Sequence by Accession Number: GenBank, RefSeq and UniProt →
• How to Identify an Unknown Sequence: DNA, RNA or Protein →