RECKONER - Read Error Corrector Based on KMC

RECKONER

RECKONER is a tool for Illumina reads correction. It is based on the algorithm BLESS, supplies indels correction and utilizes the following tools:

• KMC3 for k-mer counting,
• KMC tools for trimming a k-mer database.

RECKONER is available for Linux and Windows.

Building

Building RECKONER in Linux requires presence of Python 3 libraries (install python3-dev package in Debian or python3-devel package in Fedora). To compile use g++ 6.3.0 or newer by typing "make -j8" in the project directory. As a result executable files will be generated in a "bin" directory.

To compile RECKONER in Windows open file "RECKONER.sln" in Visual Studio 2022 or newer, select "Release" from the drop-down list. Then click "Build" and "Build Solution" from the main menu.

You can also use precompiled binaries. In Windows it could be needed to download Visual C++ 2015 Redistributable.

Running

To run RECKONER in Linux execute a file "reckoner" from the "bin" directory (typically by typing "./reckoner {ARGUMENTS} {FASTQ files}") from the command line.

To run RECKONER in Windows execute a file "reckoner.exe" from the "bin" directory (typically by typing "reckoner.exe {ARGUMENTS} {FASTQ files}") from the command line.

In both the cases the supporting tools (i.e. kmc(.exe) and kmc_tools(.exe)) have to be present in the same directory.

A list of allowed {ARGUMENTS} is given below.

• Main arguments (sufficient in a general use):
  • -help - print a short help,
  • -read - input FASTQ(.gz) read file name, can be passed many times,
  • -kmerlength - k-mers length, not mandatory,
• Supporting:
  • -genome - approximate genome size, not mandatory - when -kmerlength is not given, it can ease k-mer length determination,
• Quality tunning:
  • -longkmer - verify corrections with long k-mers (worthful for de novo assembly, requires more RAM and CPU time), it enables:
    • -accept - if it is not possible to verify some read, keep the best of its possible corrections (default: do not correct the read),
    • -longratio - long to normal k-mer length ratio, must be > 1.0,
  • -noindels - do not correct insertions and deletions (substitutions correction only),
  • -extend N - max extension length,
• Formal issues:
  • -prefix - output directory (optional, by default the current directory (.)),
  • -threads - number of correcting threads (optional, by default number of available virtual cores),
  • -mark - mark corrected reads by adding an information to their headers,
  • -fixlength - if the read headers contain read length (e.g. "length=100"), fix this value after indel correction (cannot be used with -noindels),
• Useful for debugging:
  • -kmcmemory - k-mer counting memory consumption limit in GB,
  • -kmcram - count k-mers in RAM-only mode,
  • -reuse - do not remove a KMC database; if a proper database exists, use it,
  • -verbose - return more statistics.

Instead of using -read option the input FASTQ files can be specified at the end of the command line.

K-mer length, if not given, will be determined automatically. Delivering genome size can facilitate and accelerate this determination.

To correct paired-end files simply pass both of them as two single-end files. RECKONER will keep the pairs in files.

Result files will be created in the output directory with a "corrected" suffix in the directory.

Examples:

./reckoner -kmerlength 25 -prefix results -threads 16 -read SRR088579.fastq.gz
./reckoner -read ERR729973.fastq -read ERR729974.fastq
./reckoner -kmerlength 30 ERR729973.fastq ERR729974.fastq
./reckoner -kmerlength 20 ERR422544_1.fastq.gz ERR422544_2.fastq.gz

To achieve reads best for de novo assembly:

./reckoner -kmerlength 20 -longkmer ERR422544_1.fastq.gz ERR422544_2.fastq.gz

Debugging in Windows

To debug RECKONER in Windows, select "Debug" from the Visual Studio drop-down list. In the RECKONER project properties set the working directory as:

$(SolutionDir)bin\$(Configuration)

Input files should be located in "bin\Debug".

Miscellaneous

• Contact: maciej.dlugosz@polsl.pl
• RECKONER is distributed under license GPL 3.

Citing

• Długosz, M., Deorowicz, S. (2024) Illumina reads correction: evaluation and improvements, Scientific Reports 14, 2232.
• Długosz, M., Deorowicz, S., Kokot, M. (2017) Improvements in DNA Reads Correction, International Conference on Man-Machine Interactions, 115–124.
• Długosz, M., Deorowicz, S. (2017) RECKONER: read error corrector based on KMC, Bioinformatics, 33:1086–1089.