Bio Transcription Translation

Version Compatibility

Reference examples tested with: BioPython 1.83+

Before using code patterns, verify installed versions match. If versions differ:

• Python: pip show <package> then help(module.function) to check signatures

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.

Transcription and Translation

"Translate my DNA sequence to protein" -> Transcribe DNA to RNA and translate to protein, choosing the right genetic code and validating the reading frame.

• Python: Seq.translate(), Seq.transcribe(), Bio.Data.CodonTable (BioPython)

The Governing Principle

The single most dangerous bug in translation is silent: a valid-but-wrong table= argument produces a plausible wrong protein with no error. Translating human mitochondrial DNA with the default Standard code (table 1) inserts * where UGA actually codes Trp and truncates at AGA/AGG (which are stops in vertebrate mito). The protein looks real and nothing complains. By contrast, an unknown table id or name raises KeyError (loud). Only valid-but-wrong tables corrupt silently.

The defense: when the input is a complete coding sequence, pass cds=True. It converts silent traps into loud TranslationError exceptions by validating start, length, and stop. Use it for ORF validation rather than trusting a clean-looking output.

Since the Biopython 1.78 alphabet removal, transcribe(), back_transcribe(), and translate() perform NO type checking. Transcribing a protein or translating the wrong strand returns silent garbage. Confirm the molecule and strand before converting.

Required Import

from Bio.Seq import Seq
from Bio.Data import CodonTable

Transcription Is a String Operation, Not Biology

transcribe() is a pure T->U replacement on the coding (sense) strand; back_transcribe() is U->T. Neither performs splicing, intron removal, 5' capping, or poly-A addition. The Biopython tutorial states plainly that all transcribe does is replace T with U.

coding_dna = Seq('ATGCGATCGATCG')
rna = coding_dna.transcribe()        # Seq('AUGCGAUCGAUCG'), T->U only
back = rna.back_transcribe()         # Seq('ATGCGATCGATCG'), U->T only

True biological transcription starts from the template strand, so reverse-complement first:

template = Seq('CGATCGATCGCAT')
mrna = template.reverse_complement().transcribe()

Translation accepts DNA or RNA directly, so explicit transcription is rarely needed before translate().

Translation Basics

coding_dna = Seq('ATGTTTGGT')
coding_dna.translate()               # Seq('MFG'), from DNA
Seq('AUGUUUGGU').translate()         # Seq('MFG'), from RNA

Stop-Codon Behavior

translate() substitutes stop_symbol (default '*') for EVERY in-frame stop, so internal stops appear as * mid-protein. to_stop=True instead halts at the first in-frame stop and does NOT append the symbol.

seq = Seq('ATGTTTGGTTAAGGG')
seq.translate()                      # Seq('MFG*G'), stop shown, translation continues
seq.translate(to_stop=True)          # Seq('MFG'), halts at first stop

NCBI Codon Tables: Which Code, and the Consequence of the Wrong One

Biopython exposes every NCBI genetic code by integer id or registered name. Selecting the wrong one is the #1 silent bug (see governing principle).

Explicit correction: CTG -> Thr is table 3 (Yeast Mitochondrial). Table 12 is CUG -> Ser. Do not conflate them.

seq = Seq('ATGGCCTGA')
seq.translate(table=2)                            # by NCBI integer id
seq.translate(table='Vertebrate Mitochondrial')   # by registered name

CodonTable.unambiguous_dna_by_id[2]
CodonTable.unambiguous_dna_by_name['Vertebrate Mitochondrial']

Validating a Coding Sequence with cds=True

Goal: Translate a complete ORF and have any structural defect raise a loud error instead of producing a silent wrong protein.

Approach: Pass cds=True. It enforces four conditions, each raising Bio.Data.CodonTable.TranslationError on failure: (1) first codon is a start codon for the chosen table; (2) length is a multiple of 3; (3) sequence ends in a stop; (4) no internal in-frame stop. A valid alternative start (GTG/TTG/ATT) is translated as M, biologically correct for fMet initiation. The terminal stop is stripped from the output.

Reference (BioPython 1.83+):

cds = Seq('ATGTTTGGTTAA')
cds.translate(cds=True)              # Seq('MFG'), validated, terminal stop removed

alt_start = Seq('GTGTTTGGTTAA')
alt_start.translate(table=11, cds=True)   # Seq('MFG'), GTG start -> M under bacterial code

Start-codon lists differ by table: table 1 = TTG/CTG/ATG; table 2 = ATT/ATC/ATA/ATG/GTG; table 11 = TTG/CTG/ATT/ATC/ATA/ATG/GTG. A start valid under one table fails under another, which is exactly the loud signal cds=True provides.

translate() Parameters and Edge Cases

Signature (the Seq.translate METHOD): translate(table='Standard', stop_symbol='*', to_stop=False, cds=False, gap='-'). The Seq method defaults to gap='-', while both the module-level Bio.Seq.translate(sequence, ...) function and SeqRecord.translate() default to gap=None.

• Partial codon (length not a multiple of 3): emits a BiopythonWarning and SILENTLY drops the trailing 1-2 bases. Easy to miss in a pipeline. Under cds=True the same condition becomes a loud TranslationError.
• Gaps: because the Seq method defaults to gap='-', a full gap codon '---' already translates to '-' (e.g. Seq('GTG---GCCATT').translate() -> 'V-AI', no error). A codon mixing gaps and bases ('TT-') raises TranslationError. SeqRecord.translate() instead defaults to gap=None, so even a full '---' codon raises unless gap='-' is passed; mixed gap/base codons still raise. For alignment-derived CDS, preserve codon and alignment semantics with alignment/multiple-alignment or the project's established translation wrapper rather than assuming one gap-normalization policy.
• Dual-coding stop tables (27 Karyorelict, 28 Condylostoma, 31 Blastocrithidia Nuclear): these reassign a stop codon so it codes both an amino acid and stop, so to_stop=True raises a ValueError (no single truncation point).

Seq('ATGTTTGG').translate()          # BiopythonWarning, trailing 'GG' dropped -> Seq('MF')
Seq('ATGTTTGG').translate(cds=True)  # TranslationError: length not a multiple of three