Bin3C_SLM v0.1

Deconvoluting metagenomic assemblies via Hi-C connect network

Installation

Systems requirements

• Docker
• 16GB memory recommended
• 40GB memory recommended if performing CheckM

We use Docker to build an environment for the process.

git clone https://github.com/changlabtw/Bin3C_SLM.git
cd Bin3C_SLM

# build the image 
docker build -t Bin3C_SLM . --no-cache

# run docker container, use volume to get data from host machine
docker run -it -d -v <path of data from host>:/home/vol  --name Bin3C_SLM0 Bin3C_SLM

# get a bash shell in the container
docker exec -it Bin3C_SLM0 sh 

File contents

Bin3C_SLM is based on bin3C with addtional homemade functions to perform specific clustering and evaluation.

• mzd/cluster.py: replace the original cluster.py of bin3C with two additional functions, getGraph() and getSLMresult()
  • getGraph(): convert the seq_map to an undirected Networkx Graph using to_Graph() in original cluster.py of bin3C, and generate the edge file by write_edgelist function from Networkx package.
  • getSLMresult(): combining _read_table() and part of cluster_map() function in original cluster.py of bin3C to get the sequence indices of every cluster.
• map2g.py: get an edge file from a contact map file using getGraph() in mzd/cluster.py
• SLM2seq.py: generate the sequence fasta of each bin using getSLMresult() in mzd/cluster.py
• ezcheck-full.py: evaluate the ranks (near, substantial, moderate) of the checkM result file, bin_stats_ext.tsv

Docker

• Dockerfile: for building docker image.
• requirements.txt/requirementspy3.txt: for installing required python packages during building docker image.

Tool

• bin3c_slm.sh: a wrap-up script to run the whole process of Bin3C_SLM include CheckM.

Quick start

Example data

The original dataset derives from a human fecal sample and contains a shotgun read-set (SRR6131123), and two separated Hi-C read-sets produced using two restriction enzymes MluCI (SRR6131122) and Sau3AI (SRR6131123). The following example data is generated after initial process.

• scaffolds.fasta: shotgun reads are cleaned up by BBDuk in BBTools, and assembled using metaSPAdes.
• merged_scaf.bam: merged from two bam files mapped by MluCI and Sau3AI Hi-Cs.

data download

Example usage

There are two ways to run Bin3C_SLM: one command or step-by-step.

• We supply a simple script to run the whole process include metagenome deconvolution and result evaluation.

# bin3c_slm.sh <input:assembled fasta> <input:Hi-C bam file> <output:path> <slm resolution=25.0>
bin3c_slm.sh /home/vol/data/scaffolds.fasta /home/vol/data/merged_scaf.bam /home/vol/output 25.0

• Step-by-step.
  1. generate contact map

  /home/bin3C/bin3C.py mkmap -e MluCI -e Sau3AI <input:assembled fasta> <input:Hi-C bam file> <output:path>

  2. generate connect network

  /home/bin3C/map2g.py -i <input:contact map> -o <output:path>

  3. genome binning

  java -jar /home/bin3C/external/ModularityOptimizer.jar <input:connect network> <output:path/result.txt> 1 25.0 3 10 10 9001882 1

  4. get fasta seqs based on binning

  /home/bin3C/SLM2seq.py <input:slm result> <input:contact map> <output:path>

  5. perform checkm and evaluate performance

  checkm lineage_wf -t 8 <input:fasta path>  <output:path>
  python3 /home/bin3C/ezcheck-full.py -f -i <input:bin_stats_ext.tsv from chechm> -o <output:path/ezcheck_result.csv>