AmpliconHunter2

SIMD-accelerated, in-silico PCR for very large FASTA sets. AVX2 bit-mask IUPAC matching, constant-memory streaming with 2-bit batches, multi-threaded amplicon calling with optional primer trimming and barcode extraction.

See our webserver for an easy-to-use visual interface running AmpliconHunter2 on five genomes DBs of various sizes (~5.7k genomes to ~2.4M).

Features

• AVX2 one-byte IUPAC masks with exact 3′ clamp and up to --mismatches substitutions.
• Streaming 2-bit batches and mmap-aware reads for low RSS on huge inputs.
• FR and RF by default. --include-offtarget adds FF and RR. RF is reverse-complemented on output.
• Melting temperature annotation and optional filtering (--Tm).
• Optional fixed-length flank barcodes (--fb-len, --rb-len) and --trim-primers.
• FASTA-only; see AmpliconHunter Version 1.1 if you require more advanced features.

Build

git clone https://github.com/rhowardstone/AmpliconHunter2.git
cd AmpliconHunter2; make;
sudo cp amplicon_hunter /usr/local/bin/  #for local installation

CPU must support AVX2. GCC 9+ recommended.

Quick start

1. Batch-compress FASTA to 2-bit and capture the file list for the test data:

amplicon_hunter compress --input-dir test/fasta --output test/compressed
ls -d "$PWD"/test/compressed/*.2bit > test/file_list.txt

2. Prepare primers.txt on two lines: <FORWARD>\n<REVERSE> (5' -> 3' direction), test/V1V9_primers.txt contains the following for V1V9 (full-length 16S):

AGRGTTYGATYMTGGCTCAG
RGYTACCTTGTTACGACTT

3. Run the finder:

amplicon_hunter run \
  --input test/file_list.txt \
  --primers test/V1V9_primers.txt \
  --output test/amplicons.fa

diff test/amplicons{,-check}.fa | wc -l  #should show zero

Output

FASTA with annotated headers:

>seqid.fileID=original_file.fa.source=batch_t0_b1.2bit.coordinates=12345-13567.Tm=60.54.orientation=FR.fprimer=... .rprimer=... .fb=ACGT.rb=TGCA
ACGT...

Header fields: source batch file, genomic coordinates, melting temperature, orientation, matched primer snippets, and optional barcodes.

Full syntax:

Usage:
  amplicon_hunter compress --input-dir <dir> --output <dir> [--threads <n>] [--batch-size <MB>]
  amplicon_hunter run --input <file_list> --primers <file> --output <file> [options]

Compress options:
  --input-dir <dir>    Directory containing FASTA files
  --output <dir>       Output directory for compressed files
  --threads <n>        Number of threads (default: 8)
  --batch-size <MB>    Max size per batch file in MB (default: 500)

Run options:
  --input <file>       File list from compress (single column)
  --primers <file>     Primers file (forward reverse)
  --output <file>      Output FASTA file
  --threads <n>        Number of threads (default: 8)
  --mismatches <n>     Maximum mismatches (default: 0)
  --clamp <n>          3' clamp size (default: 5)
  --min-length <n>     Minimum amplicon length (default: 50)
  --max-length <n>     Maximum amplicon length (default: 5000)
  --fb-len <n>         Forward barcode length (default: 0)
  --rb-len <n>         Reverse barcode length (default: 0)
  --trim-primers       Trim primer sequences
  --include-offtarget  Include FF and RR orientations

Tm calculation options:
  --Tm <n>             Minimum melting temperature threshold (default: 0, no filtering)
  --dnac1 <n>          Primer concentration in nM (default: 1000)
  --dnac2 <n>          Template concentration in nM (default: 25)
  --Na <n>             Sodium concentration in mM (default: 50)
  --K <n>              Potassium concentration in mM (default: 0)
  --Tris <n>           Tris buffer concentration in mM (default: 0)
  --Mg <n>             Magnesium concentration in mM (default: 4)
  --dNTPs <n>          dNTP concentration in mM (default: 1.6)
  --saltcorr <n>       Salt correction method 0-7 (default: 5)

Notes and limits

• Mismatches are substitutions only; No indels.
• Non-ACGT input symbols are mapped to the alphabetically first possible base during 2-bit packing. Uppercase and lowercase treated identically.
• Temp files are written under /tmp/amplicon_hunter_<pid> and merged automatically.
• If your input is FASTQ format, please see the Python version of our tool: AmpliconHunter Version 1.1

Benchmarking repo

For detailed usage examples, and for comparing our intermediate versions, please see our Benchmarking repo for AmpliconHunter2.

Cite

If this tool supports your work, cite the AmpliconHunter2 application note (will update with DOI when available).

License

AmpliconHunter2 is licensed under the MIT License. See LICENSE for details.