veff.py: in-frame complex variants (MNP+INDEL) return literal "TODO" string as effect — invalid scientific output

[Tanisha127]

Summary

In malariagen_data/veff.py, the function _get_within_cds_effect() has an incomplete branch for in-frame complex variants — variants that are simultaneously a multi-nucleotide polymorphism (MNP) AND an indel, where the net length change is a multiple of 3 (no frameshift).

Current broken behaviour

Lines 447-449 currently contain:

effect = base_effect._replace(
    effect="TODO in-frame complex variation (MNP + INDEL)",
    impact="UNKNOWN"
)

This means a researcher who calls snp_effects() on such a variant
receives a DataFrame where:

• The effect column contains the literal string "TODO in-frame complex variation (MNP + INDEL)"
• The impact column contains "UNKNOWN"

Why this is serious

1. "UNKNOWN" is not a valid impact level anywhere else in this codebase.
   All other values are "HIGH", "MODERATE", "LOW", or "MODIFIER".
   This silently breaks any downstream filtering such as df[df["impact"] == "HIGH"].

2. The TODO string is a developer note leaking directly into scientific output with no warning or error raised.

How to reproduce

Call snp_effects() on any variant where:

• len(ref) > 1 and len(alt) > 1 (not a simple insertion or deletion)
• len(ref) != len(alt) (not a pure MNP)
• (len(alt) - len(ref)) % 3 == 0 (in-frame, not a frameshift)

Example: ref="GCC", alt="GCCATG" at a CDS position.

Proposed fix

Replace the TODO branch with:

effect = base_effect._replace(effect="CODON_CHANGE", impact="MODERATE")

This is consistent with how pure MNPs are already handled in the elif branch directly above. Both cases represent in-frame changes to one or more codons with no frameshift.

Related

This is related to issue #1180. I will submit a PR with the fix and a regression test.