Skip to content

miFRED wiki

Jump to bottom
Sofia Fraulini edited this page Dec 16, 2025 · 3 revisions

Installation

  1. Download the miFRED folder from this repository

  2. Unzip training set files inside the Data folder

  3. Download MICROPHERRET folder from github and move it inside miFRED/Data/:

    git clone https://github.com/MetabioinfomicsLab/MICROPHERRET/

  4. Download MICROPHERRET ML models from the designated MEGA folder linked in MetabioinfomicsLab/MICROPHERRET/

    Ensure that the saved_models folder is placed inside the MICROPHERRET directory within miFRED/Data/ so that each model follows this path structure: miFRED/Data/MICROPHERRET/saved_models/

  5. miFRED requires an apposite conda environment, which can be generated as follow using the miFRED_environment.yml file in Data:

    conda env create -f miFRED_env.yml

  6. Make scripts executable using chmod +x script_name

Requirements

miFRED requires a Linux system and at least 6 CPUs due to the computational demands of the MICROPHERRET training process.

Manual

Inputs

Genomic-related inputs

  • -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDER

Directory where genomes fasta files are stored

  • -x GENOMES_EXTENSION, --genomes_extension GENOMES_EXTENSION

Genome fasta files extension (default: .fa)

  • --all_genomes ALL_GENOMES

Multi-fasta file obtained concatenating all single fasta files. Automatically generated by miFRED in input generation step by processing -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDER. Incompatible with -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDER

  • --binning_file BINNING_FILE

.txt file with each line listing a scaffold and the corresponding genome/bin name, tab-seperated. Automatically generated by miFRED in input generation step by processing -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDER with inputwriter.py additional script. Incompatible with -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDER, required if --all_genomes ALL_GENOMES is specified.

Alignment-related inputs

  • -r READS_FOLDER, --reads_folder READS_FOLDER

Directory where metagenomic reads fastq files are stored. Incompatible with -B BAM_FILES, --bam_files BAM_FILES

  • -u {True,False}, --unpaired_reads {True,False}

Required if -r READS_FOLDER, --reads_folder READS_FOLDER is specified. True if fastq file for unpaired reads are also provided in READS_FOLDER, False otherwise (default : True)

  • -B BAM_FILES, --bam_files BAM_FILES

Directory where sample-specific sorted.bam files and indexes are stored. They can be automatically generated by miFRED input generation steps aligning -r READS_FOLDER, --reads_folder READS_FOLDER against multi-fasta file (either --all_genomes ALL_GENOMES or results of -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDERprocessing). If provided check that names of aligned contigs mirror names in -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDER fasta files. If not, provide multi-fasta directly with --all_genomes ALL_GENOMES and specify right contig names in --binning_file BINNING_FILE

Functional annotation-related inputs

  • -A EGGNOG_ANNOTATION, --eggnog_annotation EGGNOG_ANNOTATION

Either directory where genomes eggNOG .annotations files are stored or .csv file obtained from parsing .annotations files, with genomes as rows, KO as columns and KO counts as values. First column must be named "Genomes"

  • -db EGGNOG_DATABASE, --eggnog_database EGGNOG_DATABASE

Directory where eggNOG-mapper database is stored, used to obtain .annotations files by launching eggNOG-mapper on genomes fasta files in the input generation steps. Incompatible with -A EGGNOG_ANNOTATION, --eggnog_annotation EGGNOG_ANNOTATION

  • -sm {default,fast,mid-sensitive,sensitive,more-sensitive,very-sensitive,ultra-sensitive}, --eggnog-sensmode {default,fast,mid-sensitive,sensitive,more-sensitive,very-sensitive,ultra-sensitive}

eggNOG-mapper Diamond search option (default: sensitive). Incompatible with -A EGGNOG_ANNOTATION, --eggnog_annotation EGGNOG_ANNOTATION

Function-related inputs

  • -f FUNCTIONS_LIST, --functions_list FUNCTIONS_LIST

.txt file containing MICROPHERRET functions to be considered for the calculation, one per line. (default: 86 functions whose models were accurate on test set, stored in functions.txt)

  • -s TRAINING_SETS, --training_sets TRAINING_SETS

Folder containing the dataset.csv and dataset_acetoclastic_methanogenesis.csv files to be used in the training. (default: ./training_sets/ )

  • -m MICROPHERRET_PREDICTIONS, --micropherret_predictions MICROPHERRET_PREDICTIONS

.csv file containing MICROPHERRET predictions for all the genomes, the first column with genomes names must be unnamed. Incompatible with -k, --KO

  • -k, --KO

Calculation must be performed using KO and not MICROPHERRET phenotypes. Incompatible with -m MICROPHERRET_PREDICTIONS, --micropherret_predictions MICROPHERRET_PREDICTIONS If both -k, --KO and -m MICROPHERRET_PREDICTIONS, --micropherret_predictions MICROPHERRET_PREDICTIONS are not specified, MICROPHERRET is launched to predict -f FUNCTIONS_LIST, --functions_list FUNCTIONS_LIST for FRED calculation.

If -k, --KO or -m MICROPHERRET_PREDICTIONS, --micropherret_predictions MICROPHERRET_PREDICTIONS are specified, -f FUNCTIONS_LIST, --functions_list FUNCTIONS_LIST and -s TRAINING_SETS, --training_sets TRAINING_SETS are ignored.

Relative abundance-related inputs

Parameters set to control calculation, aiming at excluding spurious associations.

  • -c COVERED_GENOME_FRACTION, --covered_genome_fraction COVERED_GENOME_FRACTION

Minimum fraction of genome with coverage higher than 0 (breadth of coverage) (default: 0.10)

  • -t RELATIVE_ABUNDANCE_THRESHOLD, --relative_abundance_threshold RELATIVE_ABUNDANCE_THRESHOLD

Minimum relative abundance required to consider a genome as present in a sample (default: 0)

General inputs

  • -o OUTPUT_FOLDER, --output_folder OUTPUT_FOLDER

Output directory

  • -p PROCESSORS, --processors PROCESSORS

Number of threads (default: 5)

Outputs

The following folders are generated in the output folder specified by the user.

input_fred folder

It stores the results of miFRED's input generation step procedure. The list can change depending on which files were already provided by the user.

  • all_genomes.fa

fasta file obtained by concatenating the files provided by the user. It is the reference used in the alignment procedure. Generated if -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDER is specified

  • info.txt

.txt file with each line listing a scaffold and the corresponding genome/bin name, tab-separated. Generated if -g GENOMES_FOLDER, --genomes_folder GENOMES_FOLDER is specified

  • nreads.txt

.txt file with number of reads mapped to each sample, used for relative abundance normalisation

  • bowtie2 folder

sorted and indexed bam files obtained by aligning reads against all_genomes.fa with bowtie2 and processing resulting files with Samtools. Generated if -r READS_FOLDERis specified

  • eggnog_annotations folder

contains eggNOG-mapper results. Generated if -A EGGNOG_ANNOTATION, --eggnog_annotation EGGNOG_ANNOTATION is not specified

  • annotation_matrix.csv

matrix with the genetic information (KO copy number) per genome, with genomes as rows and KO as columns; input required for prediction of phenotypes. Generated if .csv file is not already provided with -A EGGNOG_ANNOTATION, --eggnog_annotation EGGNOG_ANNOTATION

  • KO_for_calculation.csv

matrix used to calculate FRED if -k, --KO parameter is specified, with genomes as rows, KO as columns and 0/1 values to indicate presence/absence

output_micropherret folder

If MCROPHERRET models are used, it stores the results of the ML predictions:

  • predict_functions.csv

matrix with predicted functions per genome

  • predict_sum.csv

number of genomes predicted to perform each function They are generated only if -k, --KO and -m MICROPHERRET_PREDICTIONS, --micropherret_predictions MICROPHERRET_PREDICTIONS are not specified

output_fred folder

It stores results of FRED calculation procedure:

  • fredc.csv file

stores results of FREDc calculation for each analysed sample. Additional metrics like alpha diversity (Gini-Simpson index, GSI), a non-normalised version of FREDc(FREDc_tian) and Rao’s entropy for functional diversity are included too.

  • freds.csv

stores results of FREDs calculation for each function for each analysed sample

  • FREDs_statistic.csv

FREDs statistics for each analysed function and sample

Intermediate results are also provided:

  • jaccard_distances.csv

storing the pairwise functional diversity within each genome pair

  • used_relative_abundances.csv

relative abundances computed by miFRED and used for FRED calculation

  • normalised_relative_abundances.csv

relative abundances computed by miFRED, normalised by the number of mapped reads

FREDc and FREDs distribution plot are provided too to aid the analysis.

Examples of miFRED input and output files are available here: Examples

Clone this wiki locally