[BUG] MaskPrimers.py trims gapped match correctly only in the forward orientation despite correctly passing in both

[vladimirvig]

Describe the bug
When encountering an indel, the match is correctly identified. However, the primer is not properly trimmed in reverse orientation.

Steps to reproduce
Toy dataset:

>sample1
CTATCGCAAGGAACTGCCGATCTTAATGGATGGCCGGAGGTGGTATGGAAGCTATAAGCGCGGGTGAGAGGGTAATTAGGCGTGTTCACCTACGCTACGCTAACGGGCGATTCTATAAGATTGCACATTGCGTCAACTCATAAGATGTCTCAACGGCATGCGCAACTTGTGAAGTGTCTACTATCCTTAAACGCATATCTCGCACAGTAACTCCCGAATATGTCGGCATCTGATGTTGCCCGGGCCGAGTTAGTGTTGAGCTCACGGAACTTATTGTATGAGTAGTGATTTGTAAGAGTAATTATCTACTTTTTACAACAAATATAAAACAACTAGTATCGATGTCAGTTAGCTCGTTCAGGTAATAGTTGCCCACACAACGTCAAAATAAGAGAACGGTCGTAACATTATCCGTGATTTTCTCACTACTATCAGTACTCACGACTCGACTCTGCCGCAGCCACGTATCGCCTGAAAGCCAGTCAGCGTTAAGGAGTGCTCTGAGCAGGACAACTCGCGTAGTGAGAGTTACATGTTCGTTGGGCTCTTCCGACACGGACCTGAGTTGGCCAACGTCCCACCTGAGGTCTGTGCCCCGGTGATGAGAAGTGTGCATCTCGTTCTTGCAGCTCGTCAGTACTTTCAGAATCATGGCGTGCATGGTAGAATGACCCTTATAACGGACTTCGACATGGCAATAACCCCCCGTTTCTACTTCTAGAGGAGAAAAGTATTGACATGAGCGCTCCCGGCACAAGGGCCAAAGAAGTCTCCAATTTCTTATTTCCGAATGACATGCGTCTCCTTGCGGGTAAATCACCGACCGCAATTCATAGAAGCCTGGGGGAACAGATAGGTCTAATTAGCTTAAGAGAGTAAATCCTGGGATCATCGATACTAGTTGTAATTAAAACTTAGATTAGATTGCTATGCTTTCTTTCAGTAGTAACCATAAACTTACGCTGGGGCTTCTTCGGCGGATTTTTACAGTTACCAACCAGGAGATTTGAAGTAAATCAGTTGAGGATTTAGCCGCGCTATCCGGTAATCTCCAAATTAAAACATACCGTTCCATGAAGGCTAGAATTAC
>sample2
CTATCGCAAGGAACTGCCGATCTTAATGGATGGCCGGAGGTGGTATGGAAGCTATAAGCGCGGGTGAGAGGGTAATTAGGCGTGTTCACCTACGCTACGCTAACGGGCGATTCTATAAGATTGCACATTGCGTCAACTCATAAGATGTCTCAACGGCATGCGCAACTTGTGAAGTGTCTACTATCCTTAAACGCATATCTCGCACAGTAACTCCCGAATATGTCGGCATCTGATGTTGCCCGGGCCGAGTTAGTGTTGAGCTCACGGAACTTATTGTATGAGTAGTGATTTGTAAGAAGAAAGCATAGCAATCTAATCTAAGTTTTAATTACAACTAGTATCGATGTCAGTTAGCTCGTTCAGGTAATAGTTGCCCACACAACGTCAAAATAAGAGAACGGTCGTAACATTATCCGTGATTTTCTCACTACTATCAGTACTCACGACTCGACTCTGCCGCAGCCACGTATCGCCTGAAAGCCAGTCAGCGTTAAGGAGTGCTCTGAGCAGGACAACTCGCGTAGTGAGAGTTACATGTTCGTTGGGCTCTTCCGACACGGACCTGAGTTGGCCAACGTCCCACCTGAGGTCTGTGCCCCGGTGATGAGAAGTGTGCATCTCGTTCTTGCAGCTCGTCAGTACTTTCAGAATCATGGCGTGCATGGTAGAATGACCCTTATAACGGACTTCGACATGGCAATAACCCCCCGTTTCTACTTCTAGAGGAGAAAAGTATTGACATGAGCGCTCCCGGCACAAGGGCCAAAGAAGTCTCCAATTTCTTATTTCCGAATGACATGCGTCTCCTTGCGGGTAAATCACCGACCGCAATTCATAGAAGCCTGGGGGAACAGATAGGTCTAATTAGCTTAAGAGAGTAAATCCTGGGATCATCGATACTAGTTGTTTTATATTTGTTGTAAAAAGTAGATAATTACTTCAGTAGTAACCATAAACTTACGCTGGGGCTTCTTCGGCGGATTTTTACAGTTACCAACCAGGAGATTTGAAGTAAATCAGTTGAGGATTTAGCCGCGCTATCCGGTAATCTCCAAATTAAAACATACCGTTCCATGAAGGCTAGAATTAC
>sample3
CTATCGCAAGGAACTGCCGATCTTAATGGATGGCCGGAGGTGGTATGGAAGCTATAAGCGCGGGTGAGAGGGTAATTAGGCGTGTTCACCTACGCTACGCTAACGGGCGATTCTATAAGATTGCACATTGCGTCAACTCATAAGATGTCTCAACGGCATGCGCAACTTGTGAAGTGTCTACTATCCTTAAACGCATATCTCGCACAGTAACTCCCGAATATGTCGGCATCTGATGTTGCCCGGGCCGAGTTAGTGTTGAGCTCACGGAACTTATTGTATGAGTAGTGATTTGTAAGAGTAATTATCTACTTTTTACAACAAATATAAAACAACTATATCGATGTCAGTTAGCTCGTTCAGGTAATAGTTGCCCACACAACGTCAAAATAAGAGAACGGTCGTAACATTATCCGTGATTTTCTCACTACTATCAGTACTCACGACTCGACTCTGCCGCAGCCACGTATCGCCTGAAAGCCAGTCAGCGTTAAGGAGTGCTCTGAGCAGGACAACTCGCGTAGTGAGAGTTACATGTTCGTTGGGCTCTTCCGACACGGACCTGAGTTGGCCAACGTCCCACCTGAGGTCTGTGCCCCGGTGATGAGAAGTGTGCATCTCGTTCTTGCAGCTCGTCAGTACTTTCAGAATCATGGCGTGCATGGTAGAATGACCCTTATAACGGACTTCGACATGGCAATAACCCCCCGTTTCTACTTCTAGAGGAGAAAAGTATTGACATGAGCGCTCCCGGCACAAGGGCCAAAGAAGTCTCCAATTTCTTATTTCCGAATGACATGCGTCTCCTTGCGGGTAAATCACCGACCGCAATTCATAGAAGCCTGGGGGAACAGATAGGTCTAATTAGCTTAAGAGAGTAAATCCTGGGATCATCGATACTAGTTGTAATTAAAACTTAGATTAGATTGCTATGCTTTCTTTCAGTAGTAACCATAAACTTACGCTGGGGCTTCTTCGGCGGATTTTTACAGTTACCAACCAGGAGATTTGAAGTAAATCAGTTGAGGATTTAGCCGCGCTATCCGGTAATCTCCAAATTAAAACATACCGTTCCATGAAGGCTAGAATTAC
>sample4
CTATCGCAAGGAACTGCCGATCTTAATGGATGGCCGGAGGTGGTATGGAAGCTATAAGCGCGGGTGAGAGGGTAATTAGGCGTGTTCACCTACGCTACGCTAACGGGCGATTCTATAAGATTGCACATTGCGTCAACTCATAAGATGTCTCAACGGCATGCGCAACTTGTGAAGTGTCTACTATCCTTAAACGCATATCTCGCACAGTAACTCCCGAATATGTCGGCATCTGATGTTGCCCGGGCCGAGTTAGTGTTGAGCTCACGGAACTTATTGTATGAGTAGTGATTTGTAAGAAGAAAGCATAGCAATCTAATCTAAGTTTTAATTACAACTAGTATCGATGTCAGTTAGCTCGTTCAGGTAATAGTTGCCCACACAACGTCAAAATAAGAGAACGGTCGTAACATTATCCGTGATTTTCTCACTACTATCAGTACTCACGACTCGACTCTGCCGCAGCCACGTATCGCCTGAAAGCCAGTCAGCGTTAAGGAGTGCTCTGAGCAGGACAACTCGCGTAGTGAGAGTTACATGTTCGTTGGGCTCTTCCGACACGGACCTGAGTTGGCCAACGTCCCACCTGAGGTCTGTGCCCCGGTGATGAGAAGTGTGCATCTCGTTCTTGCAGCTCGTCAGTACTTTCAGAATCATGGCGTGCATGGTAGAATGACCCTTATAACGGACTTCGACATGGCAATAACCCCCCGTTTCTACTTCTAGAGGAGAAAAGTATTGACATGAGCGCTCCCGGCACAAGGGCCAAAGAAGTCTCCAATTTCTTATTTCCGAATGACATGCGTCTCCTTGCGGGTAAATCACCGACCGCAATTCATAGAAGCCTGGGGGAACAGATAGGTCTAATTAGCTTAAGAGAGTAAATCCTGGGATCATCGATACTAGTTGTTTTATATTTGTTGTAAAAAGTAGTAATTACTTCAGTAGTAACCATAAACTTACGCTGGGGCTTCTTCGGCGGATTTTTACAGTTACCAACCAGGAGATTTGAAGTAAATCAGTTGAGGATTTAGCCGCGCTATCCGGTAATCTCCAAATTAAAACATACCGTTCCATGAAGGCTAGAATTAC

primers.fasta

>PGK1
AGTAATTATCTACTTTTTACAACAAATATAAAACAACTAGTATCG
>TEF1
AAGAAAGCATAGCAATCTAATCTAAGTTTTAATTACAACTAGTATCG

Script used:

MaskPrimers.py align --mode cut -s sequence.fasta -p primer.fasta --skiprc --maxerror 0.2 --log MP1.log --failed --maxlen 800
MaskPrimers.py align --mode cut -s sequence_primers-pass.fasta -p primer.fasta --revpr --skiprc --maxerror 0.2 --log MP2.log --failed --maxlen 800

Environment and versions:

• OS: Linux
• Relevant immcantation packages and versions: pRESTO 0.7.8 (as well as 0.7.4)
• If used, the version of the Immcantation container (in the container, use builds report): immcantation/suite:4.6.0

MP1.log

MP2.log

[vladimirvig]

Particularly, sample4 has a deletion in the copy of PGK1 primer sequence, causing incorrect trimming despite correct match scoring. A similar issue is processed correctly in the forward orientation (sample3).

[vladimirvig]

(Following up using Copilot analysis... Will need to do some testing)

Steps to Reproduce

1. Run MaskPrimers.py with and without --revpr on the same sequences/primers.
2. Compare the coordinates of matched primers reported in the logs/output.
3. Observe that coordinates in reverse orientation mode do not correspond to the expected positions (for example, start coordinate can be negative or not mapped to expected bases).

Expected Behavior

• Primer coordinates, start/end, and related output fields should always correspond to their true positions relative to the original sequence, regardless of the orientation (--revpr).

Possible Solution

• In scoreAlignment, do not reassign start inside the loop over primers. Instead, use a local variable (e.g., primer_start) per primer:

if rev_primer:
    primer_start = end - len(adpt_seq)
    primer_end = end
else:
    primer_start = start
    primer_end = start + len(adpt_seq)

• Ensure that all output fields (e.g., PRSTART) are computed consistently for both orientations.

Environment

• presto version: [fill in]
• Python version: [fill in]
• OS: [fill in]

Additional Context

Original report came from troubleshooting runs where reverse orientation behaved differently.

---

Related locations:

• bin/MaskPrimers.py (scorePrimers function)
• presto/Sequence.py (scoreAlignment function)

If needed, I can help provide a patch for this issue.

[ssnn-airr]

Thanks for reporting the issue @vladimirvig and looking into it. If you already have identified the bug and possible solutions, and can add tests, it seems the most efficient way to fix this. You can open a pull request with your changes and we will review and merge when ready.

[vladimirvig]

It seems that the code in presto/Sequence.py for the localAligment() function incorrectly marks the alignment boundaries for the case when rev_primer=TRUE. This is happening despite the correct region is identified using pairwise alignment (when regex matching fails due to a gap).
The alignment coordinates used for ultimate sequence cutting/trimming/masking appear to be the same regardless of the rev_primer flag.

Will try to just use uniform case.