Skip to content

Transcript Selection

Overview

Transcript selection determines which transcript(s) are used during variant mapping, validation, and formatting.

It is one of the primary factors influencing:

  • Output size
  • Computational performance
  • Clinical relevance

All major endpoints (VariantFormatter_v2, VariantValidator, VariantValidator_ensembl, and gene2transcripts_v2) support the same transcript selection model.

Selection Modes

mane_select

Returns MANE Select transcripts only.

  • Small, highly curated dataset
  • Clinically relevant
  • One transcript per gene (where available)
  • Fastest option

Recommended default for most use cases

mane

Returns:

  • MANE Select transcripts
  • MANE Plus Clinical transcripts

This provides a broader clinically curated set than mane_select.

  • Still controlled and curated
  • Slightly slower than mane_select
  • More suitable when additional clinically relevant transcripts are required

select

Returns the canonical transcripts defined by VariantValidator.

  • Historically recommended representative transcripts
  • May differ from MANE sets
  • Moderate output size

all

Returns all available transcripts at the latest version only.

  • Large output set
  • Includes all current transcripts per gene
  • Significantly slower processing

raw

Returns all transcripts at all available versions.

  • Includes historical transcript versions
  • Produces the largest output
  • Slowest option

Explicit Transcript Selection

Users may specify one or more transcripts directly:

  • Single transcript: 

    NM_000093.4

  • Pipe-delimited string: 

    "NM_000093.4|NM_001278074.1"

  • JSON array: 

    '["NM_000093.4", "NM_001278074.1"]'

This is the most efficient method, as it bypasses transcript enumeration.

Performance Considerations

Mode Size Speed Recommendation
explicit list Minimal Fastest Best for pipelines
mane_select Small Very fast Default
mane Small–moderate Fast Clinical workflows
select Moderate Moderate Legacy compatibility
all Large Slow Use only if required
raw Very large Slowest Rare use

Recommendations

  • Use mane_select for most workflows
  • Use mane when additional clinically relevant transcripts are required
  • Use explicit transcript IDs for maximum performance
  • Avoid all and raw unless full enumeration is necessary

Notes

  • Not all genes have MANE transcripts; fallback behaviour may vary
  • Transcript availability differs between RefSeq and Ensembl
  • Transcript selection directly impacts downstream protein predictions