Read analysis

In this section we will focus on mapping metagenomic reads to AMR genes using q2-rgi.

Fetch databases¶

To run any analysis with q2-rgi, we first have to download the CARD databases with the fetch-card-db action. The action downloads and preprocesses the latest versions of the CARD, WildCARD, and kmer databases, where CARD contains curated antibiotic resistance genes, WildCARD includes predicted resistance genes from environmental genomes, and the k-mer databases provide precomputed unique k-mers for pathogen-of-origin predictions. The CARD and WildCARD databases are combined into a single artifact, while the k-mer databases are stored in separate artifacts. The CARD and WildCARD databases are used to annotate sequences with ARGs, while the k-mer database is used to predict the taxonomic origin of ARGs.

qiime rgi fetch-card-db \
    --o-card-db card_db.qza \
    --o-61-mer-db 61mer_db.qza \
    --o-15-mer-db 15mer_db.qza \
    --verbose

Analyze reads¶

To annotate reads with ARGs from CARD we can use the annotate-reads-card action. You can choose from three different alignment tools. The default and recommended aligner is KMA but Bowtie2 and BWA are also supported. Per default the action maps reads to the ARG sequences on CARD, but they can be additionally mapped to the WildCARD database and/or to three further detection models in CARD. CARD includes only curated sequences with clear experimental evidence of AMR in a peer-reviewed journal. WildCARD additionally includes in silico predicted allelic variants. It is highly recommended to use WildCARD for non-clinical samples as the allelic diversity for AMR genes is greatly unrepresented in the published literature. Please refer to the RGI documentation for more information on WildCARD and RGI. The outputs provide annotations for the allele mapping and also summarized at the AMR gene level (i.e. summing allele level results by gene). Also provided are feature tables for allele and gene mapping.

For this tutorial we will map reads to CARD and without the WildCARD. The reads used here are the same as in the end to end short tutorial.

[parsl]

[[parsl.executors]]
class = "HighThroughputExecutor"
label = "default"

[parsl.executors.provider]
class = "LocalProvider"
max_blocks = 2

qiime rgi annotate-reads-card \
    --i-reads reads.qza \
    --i-card-db card_db.qza \
    --p-aligner kma \
    --p-threads 2 \
    --o-amr-allele-annotation rgi_allele_annotations_reads.qza \
    --o-amr-gene-annotation rgi_gene_annotations_reads.qza \
    --o-allele-feature-table rgi_allele_feature_table_reads.qza \
    --o-gene-feature-table rgi_gene_feature_table_reads.qza \
    --parallel-config parallel.config.toml \
    --verbose

qiime rgi annotate-reads-card \
    --i-reads reads.qza \
    --i-card-db card_db.qza \
    --p-aligner kma \
    --p-threads 2 \
    --o-amr-allele-annotation rgi_allele_annotations_reads.qza \
    --o-amr-gene-annotation rgi_gene_annotations_reads.qza \
    --o-allele-feature-table rgi_allele_feature_table_reads.qza \
    --o-gene-feature-table rgi_gene_feature_table_reads.qza \
    --verbose

Tabulate annotations¶

With the tabulate visualizer of the q2-metadata plugin it is possible to generate a tabular combined view of the AMR annotations. The output visualization supports interactive filtering, sorting, and exporting to common file formats. All AMR annotations from q2-amrfinderplus and q2-rgi are supported to be used with the tabulate action.

qiime metadata tabulate \
    --m-input-file rgi_gene_annotations_reads.qza \
    --o-visualization rgi_gene_annotations_reads_tabulated.qzv

Your visualization should look similar to this one.

The annotations include information like the gene names their gene family, drug class and resistance mechanism. If you want to know more about a specific gene you can look up the gene in the CARD database.

Normalize feature table¶

To look at the distribution of ARGs across samples we can use the feature tables created by the annotate-reads-card action. They contain the counts of reads mapped to a specific allele or gene in each sample. These are the raw counts and have to be normalized before further analysis. Here, we will use the normalize function of the q2-feature-table plugin with the method counts per million. This removes library size biases and makes cross sample comparisons possible. It is not possible to compare the abundance of different genes within a sample because of different gene lengths. The normalisation for gene length is not yet implemented.

qiime feature-table normalize \
    --p-method cpm \
    --i-table rgi_gene_feature_table_reads.qza \
    --o-normalized-table rgi_gene_feature_table_readscpm.qza