Read quality control · Software
fastp
Quality control and preprocessing software that trims, filters and profiles short-read FASTQ sequencing data in one pass.
What fastp is
fastp is an all-in-one preprocessor and quality control tool for high-throughput sequencing data in FASTQ format. Developed by Shifu Chen and maintained by HaploX Biotechnology, it consolidates adapter trimming, quality filtering, per-read base trimming, deduplication, and quality profiling into a single execution step.
Raw sequencing data commonly contains adapter sequences, sequencing artefacts such as polyG tails from two-colour optical chemistry, low-quality base calls, and duplicate reads. Rather than piping data through multiple discrete command-line programs, fastp performs these operations in a single pass to reduce I/O overhead and processing time across short-read platforms like Illumina and MGI.
The tool produces clean, filtered FASTQ output alongside detailed quality reports in both interactive HTML and machine-readable JSON formats. These reports display quality distributions, base composition curves, GC content, duplication rates, and k-mer statistics both before and after filtering.
How it works
fastp processes sequencing reads through a defined sequence of operations: extracting unique molecular identifiers (UMIs), global front and tail trimming, sliding-window quality trimming (from 5', sliding right, or 3'), polyG and polyX trimming, adapter trimming, and trimming to a maximum length. Filter filters discard reads exceeding user-defined limits on ambiguous bases (N), average Phred quality, or the proportion of low-quality bases.
For paired-end sequencing datasets, fastp performs overlap analysis between read pairs. This overlap detection enables automatic adapter identification without pre-specified adapter sequences, correction of discordant base calls where one mate has high quality and the other low quality, and read stitching into single contiguous sequences. FASTQ-level duplication estimation and deduplication use hash algorithms to detect identical sequences.
Key facts
- Input formats
- FASTQ (single-end, paired-end, interleaved; uncompressed or gzip)
- Output formats
- FASTQ (uncompressed or gzip), HTML, JSON
- Target platforms
- Short-read sequencers (e.g. Illumina NovaSeq, NextSeq, MGI)
- Core dependencies
- isa-l, libdeflate, libhwy
What you can use it for
- Short-read quality filtering and adapter trimming
- Cleaning raw Illumina or MGI FASTQ files by removing adapter contamination and filtering low-quality or short reads prior to alignment or de novo assembly.
- Two-colour Illumina artefact removal
- Detecting and trimming synthetic polyG tails commonly generated at the ends of reads on Illumina NextSeq and NovaSeq two-colour imaging platforms.
- Paired-end read stitching and error correction
- Merging overlapping forward and reverse reads into longer single reads while resolving mismatched bases using the higher-quality mate.
- Unique molecular identifier (UMI) preprocessing
- Extracting UMIs from index sequences or read starts and moving them to the FASTQ read headers for downstream duplicate tracking in deep sequencing.
- Parallel file splitting
- Splitting processed FASTQ data into multiple chunks by read count or file number to facilitate parallel downstream alignments.
Run it here
fastp is preinstalled on the machine behind every research conversation. Ask the agent in plain language, or give it the exact command — it runs it, reads the output and explains it.
Try asking
“Clean these FASTQ files with fastp and show me the quality report.”
Strengths and limits
Good at
- Executes quality profiling, filtering, trimming, and reporting in a single pass over the sequence data.
- Identifies and trims adapter contamination in paired-end reads automatically via overlap analysis without requiring supplied adapter sequences.
- Provides comprehensive quality assessment metrics before and after filtering in both human-readable HTML and structured JSON files.
- Corrects mismatched base pairs in paired-end overlap regions when one mate has high sequencing quality and the other has poor quality.
Watch out for
- Designed specifically for short-read platforms and does not support long reads from Oxford Nanopore or PacBio instruments.
- Sliding-window quality cutting alters read lengths and coordinates, which interferes with exact-match coordinate deduplication algorithms.
- Automatic adapter detection is disabled when reading data from standard input (STDIN).
- Deduplication relies strictly on identical sequence matching and will not identify duplicate reads that contain sequencing errors or N bases.
Questions
Can I run fastp online without installing it?
fastp is preinstalled on the machine behind every research conversation. Ask the agent in plain language, or give it the exact command — it runs it, reads the output and explains it.
What licence is fastp under?
fastp is published under its authors' own licence; the project page at https://github.com/OpenGene/fastp states it.
Does fastp require adapter sequences to be specified for paired-end data?
No. For paired-end data, fastp detects adapters automatically by analysing the overlap between read pairs, although custom sequences or FASTA files can be supplied if desired.
How does fastp handle polyG tails from Illumina two-colour sequencing?
fastp detects Illumina NextSeq and NovaSeq machines from FASTQ header identifiers and automatically trims polyG tails of 10 or more bases at the 3' end by default.
Can fastp process long-read sequencing data?
fastp is built for short-read data such as Illumina and MGI; users analysing long-read data from Oxford Nanopore or PacBio are directed to the separate fastplong tool.
How are Unique Molecular Identifiers (UMIs) handled?
When UMI options are activated, fastp extracts the UMI from the specified index or read location, removes it from the read sequence if located within the read, and appends it to the read identifier for downstream alignment tools.
Can fastp merge overlapping paired-end reads?
Yes. When the merging option is enabled, overlapping read pairs meeting length and mismatch thresholds are stitched together into single reads, with the read name tagged to record the base contributions of each mate.
How to cite it
Shifu Chen. fastp 1.0: An ultra-fast all-round tool for FASTQ data quality control and preprocessing. iMeta 4.5 (2025): e70078 https://doi.org/10.1002/imt2.70078
Written from fastp's own README and project page. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.