Welcome to the MaxBin download page.

MaxBin is software for binning assembled metagenomic sequences based on an Expectation-Maximization algorithm. Users can understand the underlying bins (genomes) of the microbes in their metagenomes by simply providing assembled metagenomic sequences and the reads coverage information or sequencing reads. For users' convenience MaxBin will report genome-related statistics, including estimated completeness, GC content and genome size in the binning summary page.

Users can use MEGAN or similar software on MaxBin bins to find the taxonomy of each bin after the binning process is finished.

The most recent version of MaxBin is 2.2.5, which supports the analysis of coassemblies of multiple samples. It is available at this JBEI downloads sites as well as MaxBin and MaxBin 2.0 sourceforge sites.
For instructions on how to run MaxBin, please read the README and look at the FAQ.

MaxBin is developed under Linux platform and has been tested on docker images of six different linux distributions, including Ubuntu, Fedora, RedHat, Debian, SUSE, and Mageia. MaxBin now "semi-supports" Mac--its core program can be compiled on Mac OS X; however "./autobuild_auxiliary" is not fully supported yet. See FAQ for more details.

MaxBin is released under the BSD license.

MaxBin can also be downloaded from Sourceforge!

Docker image for the newest MaxBin version is available with repository name "karlon27/maxbin"; description on how to use the Docker image is also available at the Docker hub site.

(I would like to specially thank Upendra Kumar Devisetty, Science Analyst at CyVerse, for making a Dockerfile for creating MaxBin directly from Ubuntu Docker image (available on github) and integrating MaxBin into the CyVerse Discovery Environment tool (see here). Users who want to make MaxBin Docker image on their own are encouraged to visit the github websites.)

MaxBin publication
See below for the MaxBin manuscript supporting materials.
For MaxBin 2.0 manuscript supporting materials, please click here.

Current Version

NEW! MaxBin 2.2.7 (6/22/2019)
  • Fix the problem downloading auxiliary files

Older Versions

MaxBin 2.2.6 (1/23/2019)
  • Limit recursion up to five layers
  • Fix the version display error
  • Log will be flushed in real time

MaxBin 2.2.5 (5/8/2018)
  • Fixed the problem of "resource temporarily unavailable" when assigning threads, which will occur on newer version of g++

MaxBin 2.2.4 (7/6/2017)
  • MaxBin now supports gzipped contigs or scaffolds files.
  • Remove support for re-assembing bins due to its varied performances

MaxBin 2.2.3 (5/9/2017)
  • Fix the problem of program crash when inappropriate abundance file format is encountered.
  • Although not recommended, the delimiter for abundance files can now be either tab(\t), spaces, colons, or semicolons.

MaxBin 2.2.2 (4/13/2017)
  • Use wget instead curl to download the auxiliary software.
  • Bug fixes

MaxBin 2.2.1 (6/20/2016)
  • MaxBin will now resume from major time-consuming processes if the run aborted for some reason--users need to issue exactly the same command and keep the intermediate files to enable this feature. Note that the EM process is still treated as one whole process and cannot resumed if failed during the EM process.
  • Update FragGeneScan version to 1.30--much faster than previous versions.

MaxBin 2.2 (3/15/2016)
  • MaxBin will now collect the marker genes for each bin and make a tarball file consisting of all marker genes predicted using FragGeneScan. (I thank Richard White from Pacific Northwest National Laboratory and Alex Ranieri Jerônimo Lima from Universidade Federal do Pará, Belém, Brazil for suggesting this idea)
  • Fixed a situation that will prevent auxiliary software from executing correctly if the programs are installed under root.

MaxBin 2.1.1 (10/20/2015)
  • Fixed a bug that will show only 999 bins if the bin number reaches 1000
  • Fixed a bug that will crash MaxBin for certain format of contig header

MaxBin 2.1 (7/29/2015)
  • Change reassembly program from velvet to idba_ud.
  • autobuild_auxiliary script now records all output and error messages for the users.
  • Boost is removed from MaxBin package. MaxBin now has its own Normal and Poisson density function.

MaxBin 1.4.5 (6/12/2015)
  • Try harder finding marker genes when the number of genomes is very low in target dataset
  • Put reassembled genomes into another directory for easy comparison
  • Users can now change the minimum probability cutoff to whatever values (even 0) if they prefer
  • Fixed a bug that print EM log twice

MaxBin 1.4.4 (3/25/2015)
  • Reassembly function now supports fastq format. Users can now input interleaved fastq file for reassembly purpose.

MaxBin 1.4.3 (3/12/2015)
  • MaxBin will now try even harder to extract at least one bin from very shallow metagenome; however it is still recommended to use deeply-sequenced metagenome as MaxBin input.

MaxBin 1.4.2 (12/8/2014)
  • Fixed a bug that cannot recruit reads for reassembly.
  • Added sections instructing users to download and install auxiliary software packages if autobuild_auxiliary fails.

MaxBin 1.4.1 (11/21/2014)
  • Fixed a bug that crashes MaxBin when sequences with all 'N's are encountered.

MaxBin 1.4 (10/8/2014)
  • Added a small script "autobuild_auxiliary" to automatically download, build, and setup auxiliary software packages from mirror sites. This should reduce most of the effort from building and setting auxiliary software.
  • Added -marketset option to choose between 107 marker genes and 40 marker genes sets. 40 marker genes are more suitale for extreme environments that dominated by Archaea while 107 marker genes are suitable for more general cases. Default is 107 marker genes.
  • Added -reassembly parameters for recruiting reads and then reassemble individual bins. This feature is still highly experimental. Feedbacks are more than welcomed.
  • Added Velvet as another mandatory auxiliary software package.
  • Due to the newly-released FragGeneScan 1.19 (which I helped to develop) that supports multi-thread function, MaxBin will now attempt to run FragGeneScan in multi-thread mode.

MaxBin 1.3 (5/12/2014)
  • Fixed bugs relating to deal with 100+ bins.
  • Cleaned up the log file.
  • Added the -max_iteration option to adjust the maximum number of EM runs (default 50).
  • Unclassified sequences and sequences shorter than minimum length threshold are now stored in different files (.tooshort and .noclass).

MaxBin 1.2.1 (3/14/2014)
  • Added support for the FASTQ format when the -reads option is specified.

MaxBin 1.2 (1/27/2014)
  • Stabalized the connection between MaxBin and auxiliary software packages.
  • Added the function to check the output of auxiliary software packages. MaxBin will now output warning messages and stop running when there are errors.
  • Bug fixes.

MaxBin 1.1 (10/15/2013)
  • Fixed the bug that crashes the program when a FASTA file with a long single line is encountered.
  • Implemented recursive binning, in which binned FASTA files will be checked separately for further binning possibilities.
  • Make the summary file more informative by adding Completeness (i.e. how complete the binned genomes are), Genome sizes and GC content.
  • Sort the bins by descending order of their abundance levels.
  • Add total number of marker genes and the number of unique marker genes into 2nd and 3rd column of the .marker file.
  • Add a function to automatically "switch" to more powerful binning mode once MaxBin finds that it cannot find enough marker genes to determine bin number.

MaxBin 1.0.1 (6/25/2013)
  • Added -min_contig_len parameter for users to decide the minimum contig length acceptable in MaxBin. Default value of minimum contig length is 1000.
  • Added -plotmarker option and integrated it with R (Rscript) to plot the marker gene heatmap as PDF file.
  • Added -verbose option for debug usage.
  • Fix bug when the coverage of certain contig is 0. This should not happen very often, but codes are added to avoid program crash on such circumstance.

MaxBin 1.0 (not available for download) (4/29/2013)
  • Integrated core MaxBin algorithm into perl script to connect with 3rd-party software.
  • Added compile_MaxBin script for users to compile MaxBin core program and all 3rd-party software.

MaxBin Manuscript Supporting Materials

MaxBin 2.0 Manuscript Supporting Materials

  • Click here for the contigs of co-assembled 37A and 37B samples, and here for the binning result
  • Click here for the contigs of the co-assembly of ten HMP samples, and here for the binning result.

MaxBin FAQ

Q: How do I compile MaxBin from scratch?
A: Please run make under src directory to build the core program. You also need to download and install auxiliary software packages, including Bowtie2, FragGeneScan, HMMER3, and IDBA to run MaxBin correctly. Run the script autobuild_auxiliary to automatically download, build, and set all auxiliary software packages.

Q: Is MaxBin compatible with my linux version?
A: MaxBin was tested on docker images of six different linux distributions to ensure that MaxBin is compatible with all of them. If you cannot install MaxBin on your linux using simply "make" and "./autobuild_auxiliary," please contact me. I will work out a solution for you.

Q: Is MaxBin compatible with my mac?
A: MaxBin core program can be compiled on mac OS X with single thread support; however the "./autobuild_auxiliary" script cannot be used to automatically download and compile the auxiliary software packages. Users wish to use MaxBin on mac needs to download, install, and setup the software by themselves.

Q: I already have my own auxiliary software packages installed in my computer. Do I need to install them again?
A: Technically you don't need to install the software packages again; however, if your software packages are older than that used in MaxBin test, we recommend you update or download and install new software packages again (or simply run autobuild_auxiliary script). This will ensure MaxBin running process is free of errors.

Q: I don't have abundance information for my assembled scaffolds. What can I do?
A: MaxBin can find the abundance information from the sequencing reads automatically. Simply provide your reads file (in either FASTQ or FASTA format) using the -reads option. Multiple reads files can be provided separately using parameters -reads, -reads2, -reads3, ..., up to the number of reads that you wish to provide.

Q: I ran MaxBin with -reads option. Where can I find the scaffold abundance information that MaxBin generated for me?
A: The abundance information is stored at (output file header).abund. You can use this file for re-run MaxBin for any other runs.

Q: I have a few metagenomic reads (or abundance) files to feed MaxBin. How should I specify the parameters?
A: Please use -reads, -reads2, reads3, or -abund, -abund2, -abund3, etc. to specify the filenames. You can use a combination of these parameters to specify your metagenomic reads or abundance files.

Q: I have a lot of reads (or abundance) files to feed MaxBin. Is there a better way than simply typing all names?
A: You can list all reads or abundance files (with relative or absolute paths) in a file, one file per line, and use parameter -reads_list or -abund_list.

Q: Can I feed both reads and abundance files to MaxBin?
A: Yes you can use a combination of -reads, -abund, -reads_list, and -abund_list as parameters for using MaxBin. However, please note that do NOT feed reads and abundance files that BELONG TO THE SAME METAGENOME to MaxBin. This may result in skewed results since the metagenome will be accounted for TWICE!

Q: I have a question about MaxBin. Who should I contact?
A: Please contact the developer Yu-Wei Wu at ywwei@lbl.gov.