Shotgun Metagenomics

Unlike 16S rRNA amplicon sequencing which targets a single marker gene, shotgun metagenomics fragments and sequences all genomic DNA present in a sample. This includes bacterial, archaeal, viral, fungal, and host DNA without prior selection. The process involves mechanical or enzymatic shearing of extracted DNA into small fragments, adapter ligation, and high-throughput sequencing on platforms like Illumina NovaSeq. The resulting short reads represent a random sampling of the entire metagenomic pool, enabling simultaneous analysis of taxonomic composition, functional potential, and novel genome discovery from a single sequencing run.

A critical advantage over marker gene approaches is the ability to profile the collective metabolic potential of a microbial community. Sequenced reads are aligned against curated functional databases to annotate protein-coding genes and pathways:

• KEGG Orthology (KO): Maps genes to metabolic pathways and modules
• eggNOG: Assigns orthologous groups and functional categories
• CAZy: Identifies carbohydrate-active enzymes for biomass degradation studies
• CARD and ResFinder: Detect antimicrobial resistance determinants This functional profiling reveals not just who is there, but what they are capable of doing—essential for understanding biogeochemical cycling, host-microbe interactions, and disease mechanisms.

Shotgun data enables the computational reconstruction of Metagenome-Assembled Genomes (MAGs) from uncultivated microorganisms. Reads are first assembled into contiguous sequences (contigs) using metagenomic assemblers like metaSPAdes or MEGAHIT, which handle non-uniform coverage distributions. Contigs are then binned into discrete population genomes based on tetranucleotide frequency and differential coverage across samples. Tools like CheckM assess genome completeness and contamination using lineage-specific single-copy marker genes. This approach has recovered thousands of novel phyla from environments ranging from the human gut to deep-sea hydrothermal vents.

Shotgun metagenomics provides the sequencing depth and genomic coverage necessary to resolve sub-species genetic variation. By aligning reads against reference genomes or analyzing single nucleotide variant (SNV) patterns across samples, tools like StrainPhlAn and inStrain can distinguish closely related strains within the same species. This resolution is critical for:

• Pathogen outbreak tracking: Identifying transmission chains from single-nucleotide differences
• Microbiome engraftment studies: Tracking donor strains in fecal microbiota transplantation
• Functional strain heterogeneity: Linking specific genetic variants to phenotypic differences in metabolism or virulence

Because shotgun metagenomics captures host DNA alongside microbial DNA, it enables simultaneous analysis of both compartments from a single sample. In clinical metagenomics, this dual-readout supports:

• Host transcriptome profiling: Aligning RNA-seq reads to the host genome to assess immune response gene expression
• Host genetic variant calling: Identifying human leukocyte antigen (HLA) types or immune-related polymorphisms
• Microbial load quantification: Calculating the ratio of microbial to host reads as a proxy for infection burden Computational host depletion tools like KneadData and Bowtie 2 subtract host reads before microbial analysis, while preserving them for parallel host-focused investigations.

Unlike 16S sequencing which is blind to viruses, shotgun metagenomics captures DNA and RNA viral sequences as well as extrachromosomal mobile genetic elements. Specialized tools identify these entities through distinct genomic signatures:

• VirSorter2: Uses a random forest classifier trained on viral hallmark genes and genomic features to detect diverse bacteriophage and eukaryotic viral sequences
• geNomad: Simultaneously identifies plasmids and viruses using a neural network that evaluates gene content and synteny
• PlasFlow: Employs a deep learning model to distinguish plasmid contigs from chromosomal contigs based on k-mer frequency signatures This capability is essential for studying phage-host dynamics, horizontal gene transfer, and the environmental virome.

The computational process of grouping assembled contiguous DNA sequences (contigs) into discrete population genomes, known as Metagenome-Assembled Genomes (MAGs). Binning algorithms leverage two primary signals: sequence composition (tetranucleotide frequency) and differential coverage across multiple samples. Tools like MetaBAT2 and CONCOCT use these features to cluster contigs that likely originate from the same organism, enabling the recovery of genomes from uncultivated taxa.

The computational characterization of microbial community structure by identifying and quantifying constituent organisms at various taxonomic ranks. Profilers fall into two categories: read-based tools like Kraken2 and MetaPhlAn that classify individual reads against reference databases, and assembly-based methods that bin contigs into MAGs. Output is a relative abundance table, often visualized as a sankey diagram or stacked bar chart.

The frequency distribution of all possible nucleotide subsequences of a fixed length k within a sequencing read or genome. The k-mer spectrum serves as a fundamental compositional feature for assembly-free classification. Key applications include:

• Genome size estimation from peak coverage
• Error correction by identifying low-frequency erroneous k-mers
• Contamination detection via anomalous k-mer profiles

The computational process of characterizing the collective metabolic and functional potential encoded by a metagenome. This involves mapping sequencing reads or predicted genes to curated orthologous databases such as KEGG Orthology, eggNOG, or Pfam. Tools like HUMAnN 3.0 stratify functional contributions by contributing species, enabling the linkage of specific metabolic pathways to individual community members.

The analytical capability to distinguish genetic variants below the species rank, critical for pathogen outbreak tracking and understanding functional differences within a microbial population. Achieved through:

• Single nucleotide variant (SNV) profiling across core genes
• Pan-genome analysis to identify strain-specific accessory genes
• Tools like StrainPhlAn and inStrain that use consensus and polymorphic site patterns

The bioinformatic process of identifying known or novel genetic determinants of antibiotic resistance from metagenomic data. Reads or assembled contigs are aligned against curated databases like the CARD (Comprehensive Antibiotic Resistance Database) or ResFinder. Deep learning models like DeepARG now augment alignment-based methods to detect remote homologs and predict resistance phenotypes directly from sequence context.