An ultra-fast all-in-one FASTQ preprocessor (QC/adapters/trimming/filtering/splitting/merging...)
Fix the bug in the mode of interleaved input
The threading and I/O modules have been completely rewritten to generate reproducible outputs and improve performance greatly. New libraries libisal
and libdeflate
were introduced to replace the slow zlib
. Although this may bring some difficulties to compilation, it is all worthwhile for performance improvement.
In many cases, fastp v0.23.0 can be 2x as faster as previous versions. Especially, when the compression level is set to 6 or higher, the performance gain is very obvious.
Threading randomness issue has been addressed, so the output files are MD5 consistent when you run it twice, which means the results are completely reproducible.
Support deduplication (--dedup), and refine the duplication evaluation algorithm.
https://cdn.plot.ly is not available from some places, and it will block fastp's HTML figures. From this release, we will use opengene.org to distribute plotly.js.
Revise overlap detection, PE correction and adapter trimming Support average quality score filter Count polyX ...
new -m/--merge
option to merge paired reads.