Analyzing metagenomes with QIIME

Because the metagenomes are provided in BIOM format by default, these can be plugged into many of the downstream analysis tools available in QIIME. QIIME’s Shotgun Metagenome Analysis tutorial illustrates a couple of the steps that can be applied. The steps that will primarily be of interest in that tutorial are the ones that begin with a .biom file. For example, computing beta diversity and PCoA plots and generating summaries of samples by KO categories.

Many of QIIME’s tutorials that describe diversity analyses are applicable to PICRUSt-predicted metagenome tables. Specific analysis tools that may be useful include:

• alpha_diversity.py
• beta_diversity.py
• compute_core_microbiome.py
• jackknifed_beta_diversity.py
• make_distance_boxplots.py
• alpha_rarefaction.py
• beta_diversity_through_plots.py
• group_significance.py
• shared_phylotypes.py

Plots of functional categories at various levels can be created using summarize_taxa_through_plots.py

• Since KEGG Orthologs belong to several pathways you should collapse your PICRUSt predictions to the desired hierarchy level using categorize_by_function.py

  categorize_by_function.py -i metagenome_predictions.biom -c "KEGG_Pathways" -l 2 -o metagenome_at_level2.biom
  

• Then add the following lines to a qiime parameter file (e.g. qiime_params.txt) ensuring that the level you collapsed at is the same in your config file

  summarize_taxa:md_identifier    "KEGG_Pathways"
  summarize_taxa:absolute_abundance   True
  summarize_taxa:level    2
  

• Lastly, run summarize_taxa_through_plots.py

  summarize_taxa_through_plots.py -i metagenome_at_level2.biom -p qiime_params.txt -o plots_at_level2
  

There are also a number of scripts in QIIME that may be useful for more general processing of your BIOM table. These include the following:

• single_rarefaction.py
• filter_otus_from_otu_table.py
• filter_samples_from_otu_table.py
• per_library_stats.py
• filter_taxa_from_otu_table.py
• merge_otu_tables.py
• sort_otu_table.py
• split_otu_table.py
• split_otu_table_by_taxonomy.py

Note that while many of these refer to OTU table, it’s just a nomenclature issue. These are generally applicable to .biom tables.

Finally, if you’re interested in comparing real to predicted metagenomes, or predicted metagenomes to 16S, you’ll be interested in the Procrustes Analysis tutorial and the Comparing Distance Matrices tutorial.