Incorrect adapter trimming in single-end mode for amplicon sequencing

[lynxed]

Hi,
I encountered an issue when using fastp on amplicon sequencing data (insert size ~200 bp, paired-end 150 bp).

When I run fastp with only R1 input (single-end mode), fastp mistakenly identifies the first 60bp of the 5′ of the reads as adapters, and trims them off.
As a result, more than 90% of reads are lost.

This issue does not occur when both R1 and R2 are provided in paired-end mode.

command used:
fastp -i r1.fq.gz -o clean.fq.gz -w 4 -j fp.fastp.json

fastp version:
0.22.0 and 1.0.1

Example of wrong trimming behavior (from JSON report):
"adapter_cutting": { "adapter_trimmed_reads": 76224, "adapter_trimmed_bases": 11264918, "read1_adapter_sequence": "TGTGGCAGAAGAACATAGCAGGAGGTTGAGTTTCCATTCCGCCAGCTACAGCAGGAAGGC", "read1_adapter_counts": {"TGTGGCAGAAGAACATAGCAGGAGGTTGAGTTTCCATTCCGCCAGCTACAGCAGGAAGG":1078, "TGTGGCAGAAGAACATAGCAGGAGGTTGAGTTTCCATTCCGCCAGCTACAGCAGGAAGGCACAGTGAGGGTGCCTAAGATTCTGGTAGAATCTGACCACCAGAGGATGGAGCCCAGAGACAACCACTCCCAGGGGAAAGGAAGGGGAATC":67975, "TGTGGCAGAAGAACATAGCAGGAGGTTGAGTTTCCATTCCGCCAGCTACAGCAGGAAGGCACAGTGAGGGTGCCTAAGATTCTGGTAGAATCTGACCACTAGAGGATGGAGCCCAGAGACAACCACTCCCAGGGGAAAGGAAGGGGAATC":928, "others":6243} }

Thanks a lot for your time and for maintaining this excellent tool!