bfqutils: Ben's FastQ Utilities

A set of utilties available as subcommands within a single program, bfqutils.

• FastQ PE read merging: bfqutils merge
• FastQ adapter trimming for paired-end reads: bfqutils trimpe
• FastQ trimming for single-end reads: bfqutils trimse
• FastQ statistics: bfqutils stats

Installation

Requires gcc/clang and GNU Make, tested with macOS and Linux. bfqutils is written in C and comes bundled with the Cloudflare fork of Zlib.

git clone https://github.com/noborilab/bfqutils
cd bfqutils
make libz 
make release
make test     # optional

To dynamically link to a system Zlib instead of building a static library:

make z_dyn=1 release

bfqmerge

This tool is meant to be used as a drop-in replacement for fastp --merge. The only differences from fastp are a smaller default minimum merge length (since this tool is primarily intended for libraries with small inserts) and more aggressive base correction. Apart from that, bfqmerge is intended to be a lightweight replacement with extremely low memory usage (<2 MB) and very good single-thread performance (<15 min on 2x100M PE150 reads, ~10 min without the -z flag). Remember to change the default values in src/bfqmerge.c if that would better suit your primary use case.

Usage

bfqutils v1.3.0  Copyright (C) 2026  Benjamin Jean-Marie Tremblay
Usage:  bfqutils merge [options] R1.fq[.gz] R2.fq[.gz] > merged.fq
 -o <int>  Required overlap for a merge to occur. Default: 15
 -d <int>  Maximum number of mismatches between alignments. Default: 5
 -p <dbl>  Maximum fraction of mismatches between alignments. Default: 0.2
 -Q <int>  Minimum PHRED+33 quality to consider a base high quality. Default: 15
 -u <dbl>  Maximum fraction of bases allowed to be low quality. Default: 0.4
 -n <int>  Maximum number of Ns allowed. Default: 5
 -g <int>  Number of Gs to trigger polyG tail trimming. Default: 10
 -t <int>  Mean window quality threshold for trimming 3-prime bases. Default: 20.
 -w <int>  Window size of 3-prime base trimming. Default: 5
 -m <int>  Max merged read length.
 -z        Compress the output as gzip.
 -q        Make the program quiet.
 -v        Print the version and exit.
 -h        Print this help message and exit.

Implementation details

bfqmerge is a simple program, which performs the following operations:

1. Detect and trim low quality bases from the 3' end of the reads (-t, -w). This helps reduce the possibility of false positive alignments between incorrect read segments, especially when expecting short inserts.

2. Detect and trim polyG tails (-g). Failure to remove these can lead to bfqmerge thinking that sufficiently long stretches of Gs found in both reads of a pair are the overlapping parts of the reads. Due to the way some machines work, sequencing past the end of a read can lead to long stretches of Gs with high quality scores (which won't be trimmed in step 1).

3. Find the best overlap, depending on user settings (-o, -d, -p).

4. If no overlap is found, discard the reads. Otherwise, create a new merged read. For each overlapping position, the base and quality score is taken from whichever of the two reads has a better score in that position.

5. Check if the merged read passes quality filters (-Q, -u, -n). If yes, then compress the output if desired (-z), then write to stdout.

With the exception of -o, all default values are identical to fastp. These generally work quite well, though be aware that it is impossible to get read merging right 100% of the time. False positive merging events and false negative read discards will almost always occur for any combination of settings given a sufficiently diverse set of reads.

bfqtrimpe

Paired-end adapter trimming using overlap analysis (no adapter sequence required). Like bfqmerge, it finds the overlap between R1 and R2 to locate the insert boundary, then truncates each mate to the insert length instead of collapsing them into one read. When no overlap is found, the pair passes through with only 3′ quality trimming and polyG trimming applied. Both mates are filtered independently; if one fails the quality filter its surviving mate can be written to a separate singletons file (-s).

Usage

bfqutils v1.3.0  Copyright (C) 2026  Benjamin Jean-Marie Tremblay
Usage:  bfqutils trimpe [options] R1.fq[.gz] R2.fq[.gz]
 -o <int>   Required overlap to detect adapters. Default: 15
 -d <int>   Maximum number of mismatches in the overlap. Default: 5
 -p <dbl>   Maximum fraction of mismatches in the overlap. Default: 0.2
 -1 <str>   Output file for trimmed R1 reads.
 -2 <str>   Output file for trimmed R2 reads.
 -i         Interleaved output to stdout. Mutually exclusive with -1/-2.
 -s <str>   Output file for singleton reads (surviving mate of a failed pair).
 -Q <int>   Minimum PHRED+33 quality to consider a base high quality. Default: 15
 -u <dbl>   Maximum fraction of bases allowed to be low quality. Default: 0.4
 -n <int>   Maximum number of Ns allowed. Default: 5
 -g <int>   Number of Gs to trigger polyG tail trimming. Default: 10
 -t <int>   Mean window quality threshold for trimming 3-prime bases. Default: 20.
 -w <int>   Window size of 3-prime base trimming. Default: 5
 -m <int>   Max read length.
 -M <int>   Min read length. Default: 15
 -z         Compress the output as gzip.
 -q         Make the program quiet.
 -v         Print the version and exit.
 -h         Print this help message and exit.

Implementation details

bfqtrimpe performs the following operations on each read pair:

1. Detect and trim low quality bases from the 3′ end of both mates (-t, -w).

2. Detect and trim polyG tails from both mates (-g).

3. Find the best overlap between R1 and R2 using the same algorithm as bfqmerge (-o, -d, -p). If an overlap is found, both mates are trimmed to the insert length (the region before the adapter read-through begins). If no overlap is found, both mates are kept at their current lengths.

4. Check each mate independently against the quality filters (-Q, -u, -n, -M, -m). If both pass, write the pair to the -1/-2 output files (or interleaved stdout with -i). If one mate passes and one fails, write the surviving mate to the singletons file (-s) if specified, otherwise drop both.

bfqtrimse

A simple FastQ trimming and quality filtering tool for single-end reads. The order of operations is similar to bfqmerge, replacing overlap analysis with adapter sequence matching. Use '-' for stdin.

Usage

bfqutils v1.3.0  Copyright (C) 2026  Benjamin Jean-Marie Tremblay
Usage:  bfqutils trimse [options] reads.fq[.gz] > trimmed.fq
 -a <str>   Adapter sequence. Default: AGATCGGAAGAGCACACGTCTGAACTCCAGTCA
 -Q <int>   Minimum PHRED+33 quality to consider a base high quality. Default: 15
 -u <dbl>   Maximum fraction of bases allowed to be low quality. Default: 0.4
 -n <int>   Maximum number of Ns allowed. Default: 5
 -g <int>   Number of Gs to trigger polyG tail trimming. Default: 10
 -t <int>   Mean window quality threshold for trimming 3-prime bases. Default: 20.
 -w <int>   Window size of 3-prime base trimming. Default: 5
 -M <int>   Min trimmed read length. Default: 15
 -m <int>   Max trimmed read length.
 -z         Compress the output as gzip.
 -q         Make the program quiet.
 -v         Print the version and exit.
 -h         Print this help message and exit.

bfqstats

Meant to be used in a pipe with bfqmerge or bfqtrimse. The default top enriched K-mer summary can help spot bad trimming/merging. Use '-' for stdin.

Usage

bfqutils v1.3.0  Copyright (C) 2026  Benjamin Jean-Marie Tremblay
Usage:  bfqutils stats [options] reads.fq[.gz]
 -l <file>  Read Length histogram.
 -g <file>  Read GC content histogram.
 -q <file>  Mean read quality histogram.
 -Q <file>  Per-position mean quality histogram.
 -b <file>  Per-position base content histogram.
 -k <file>  K-mer counts and obs/exp ratios.
 -o <file>  Send summary stats to a file instead of stderr.
 -K <int>   K-mer size for -k. Default: 6
 -n <int>   Only examine this number of reads.
 -O         Send the reads to stdout.
 -z         If -O, compress as gzip.
 -N         Do not print summary stats to stderr.
 -v         Print the version and exit.
 -h         Print this help message and exit.