Skip to content

Analyzing metagenomes for Antimicrobial Resistance (AMR) Genes

We're going to use AMRFinderPlus, together with the megahit metagenome assembler and the prodigal gene finder, to look for antimicrobial resistance genes in the CD136 metagenome.

We're going to do this by assembling the CD136 metagenome using the megahit assembler. This will give us contigs that represent the high coverage portion of the metagenome.

Install amrfinder, megahit, and prodigal.

First, install the software. Run: 

mamba create -n amrfinder -y ncbi-amrfinderplus megahit prodigal csvtk matplotlib seaborn pandas
conda activate amrfinder

Next, download the amrfinderplus database. Run: 

amrfinder -u

And, finally, set up & change to a working directory. Run: 

mkdir ~/amr/
cd ~/amr/

Assemble your metagenome:

We'll start by assembling the CD136 metagenome into contigs. In this case, we're not going to bin the contigs, because AMR genes don't assemble well, and in particular don't assemble into regions that are connected to their host genome. So we run the assembler, and look at genes on the resulting contigs.

Run: 

megahit -1 ../data/IBD_tutorial_subset/metag/1-trimmed/CD136/CD136.1.paired.fq.gz \
    -2 ../data/IBD_tutorial_subset/metag/1-trimmed/CD136/CD136.2.paired.fq.gz \
    -o CD136.assembly.d
This will take about 2 minutes, and produce (among other files) a set of assembled contigs in CD136.assembly.d/final.contigs.fa.

Now, predict proteins in the assembled contigs using prodigal: 

prodigal -p meta -q -i CD136.assembly.d/final.contigs.fa -a CD136.assembly.faa

This will produce a FASTA file containing many protein sequences: 

head CD136.assembly.faa
These are the (partial & complete) genes found by the prodigal software.

And, finally, run AMRfinder on the proteins: 

amrfinder -p CD136.assembly.faa -t 16 -o CD136.amrfinder.tsv --plus
(This will take under a minute.)

AMRfinder will produce a spreadsheet named CD136.amrfinder.tsv that contains a number of columns - you can see the list like so, using csvtk headers:

csvtk -t headers CD136.amrfinder.tsv

To pick out just a few columns, you can use csvtk cut.

Run: 

csvtk -t cut -f "% Coverage of reference sequence","HMM description" CD136.amrfinder.tsv

and you will see: 

% Coverage of reference sequence        HMM description
89.41   CfxA family broad-spectrum class A beta-lactamase
87.59   23S ribosomal RNA methyltransferase Erm
52.84   NA
100.00  macrolide efflux MFS transporter Mef(En2)
100.00  lincosamide nucleotidyltransferase Lnu(AN2)
100.00  CepA family extended-spectrum class A beta-lactamase

The first column here is the amount of the known (reference) sequence that is present in the metagenome, and the second is the description of the match.

Note: If you wanted to get the abundance of these in the metagenome, you would have to find the DNA contig that the relevant gene was on, using the column "Protein identifier", and then map the metagenome reads to it to get the abundance. This is because assembly collapses the abundance of the output contigs, and you have to recover it through other means.

Now we can plot the abundances of the AMRs we found: 

../scripts/plot_amr.py CD136.amrfinder.tsv -o CD136.amrfinder.pdf
CD136 amrfinder