PR checklist

• This comment contains a description of changes (with reason).
• If you've fixed a bug or added code that should be tested, add tests
• Ensure the test suite passes (nextflow run . -profile test,docker --outdir <OUTDIR>).
• Check for unexpected warnings in debug mode (nextflow run . -profile debug,test,docker --outdir <OUTDIR>).

Description of changes

• Protein download improvements:

  • Filtered out assemblies that do not contain sequences to prevent errors during downstream processing (with debug log prints).
  • Added retry logic for Entrez handles to make downloads more robust against intermittent network or server errors.
  • Protein chunks are now correctly appended rather than overwriting previous chunks in the protein list -> Fixed protein chunk overwriting.

• Peptide generation improvements:

  • Debug logging now prints all proteins containing invalid amino acids (determined with extended amino acid alphabet).
  • Filtered out all peptides that do not match the basic amino acid alphabet to ensure compatibility with MHCflurry and maintain standardisation across prediction tools.

Tests

• Pipeline tested with: nextflow run . -profile debug,test,m3 --outdir output without unexpected warnings
• Pipeline tested on a test condition with 10 Taxa that trigger all added different cases
• Pipeline tested on a bigger dataset of around 150 taxa (For testing protein download and epitope prediction parallelisation)