/MetaBIDx

A new computational tool to empowering bacteria identification in microbiomes

Primary LanguageGoMIT LicenseMIT

MetaBIDx

Description

MetaBIDx is a novel identification tool that utilizes a modified Bloom filter for efficient indexing of reference genomes and incorporates a strategy for reducing false positives by clustering species based on their genomic coverages by classified reads.

MetaBIDx requires a set of reference genomes (FASTA format) for building an index. For classifying reads or identifying species, the tool needs MetaBIDx index and metagenomes (FASTQ format) as inputs.

In this repo, we provide an example of a set of reference genomes in test_data/references which can be used to build MetaBIDx index. We also provide an example of a pair of Fastq files in test_data/Reads, which can be used as an input in querying reads or predicting species.

Download MetaBIDx

Index genomes in a microbiome

metabidx build -refseq genome_folder -save index_name.bin -k 15

where:

  • genome_folder: Path to reference sequence genome directory
  • index_name: prefix of an output index
  • k: k-mer length (Default: 31)

For more optional parameters, run metabidx build -h

Example:

metabidx build -refseq test_data/Two_refs/ -save two_refs_index.bin

Query reads

metabidx query -load path/to/index_name.bin -r1 path/to/read_1.fq -r2 path/to/read_2.fq -out query_output.txt

where:

  • index_name is the name of the index
  • read_1 is read or the first read in pairs
  • read_2 is the second read in pairs. This input is optional.
  • query_output.txt is the file containing output of classified reads. (Default: query_output.txt)

For more optional parameters, run metabidx query -h

Example:

metabidx query -load my_index.bin -r1 test_data/Reads/r1.fq -r2 test_data/Reads/r2.fq -out my_query_output.txt

Query output format

  • The query output has 2 columns:
    • First column: read header
    • Second column: the species name which the read is classified to or "unclassified"

Predict species

Requirements

  • Python3 (tested with Python v3.11.4)
  • Python3 package: scikit-learn, yellowbrick, pandas, numpy
    • Install via conda
     conda install -y anaconda::pandas
 conda install -y anaconda::scikit-learn
 conda install -y conda-forge::yellowbrick
 conda install -y anaconda::numpy
    • Install via pip
     pip install pandas
 pip install -U scikit-learn
 pip install yellowbrick
 pip install numpy

Predict species

metabidx predict -load path/to/index_name.bin -r1 path/to/read_1.fq -r2 path/to/read_2.fq -out prediction_output.txt -python-path /home/user/miniconda3/bin/python3

where:

  • index_name is the name of the index
  • read_1 is read or the first read in pairs
  • read_2 is the second read in pairs. This input is optional.
  • prediction_outputs.txt is the file containing predicted species names. (Default: prediction_output.txt)

For more optional parameters, run metabidx predict -h

Example:

metabidx predict -load my_index.bin -r1 test_data/Reads/r1.fq -r2 test_data/Reads/r2.fq -out my_prediction_output.txt

Predict output format

  • The predict output file contains all predicted species names.