Background Material¶

Pre-Processing¶

Types of Analysis¶

Comparative Analyses¶

Learning Objectives¶

• Assess the overall quality of NGS (FastQ format) sequence reads
• Visualise the quality, and other associated matrices, of reads to decide on filters and cutoffs for cleaning up data ready for downstream analysis
• Clean up adaptors and pre-process the sequence data for further analysis

Examples of Tutorial with Specific Learning Objectives¶

crAssphage Assembly Workshop

Marker gene amplicons¶

• 16s rRNA, 18s rRNA, ITS1/2
• Taxonomic profiling
• OTU clustering
• taxonomy -> function inferences

Links to resources that cover several for the above topics¶

• Hsiao tutorials
• Morgan Langille
• EDAMAME course

Learning objectives¶

• Define a marker gene and its taxonomic relevance.
• Distinguish between amplicon vs. shotgun approach.
• Apply a marker gene analysis workflow
• Analyse the taxonomic composition of an environment.
• Interpret results in biological context
• Describe and execute an amplicon workflow
• Install software for amplicon analysis
• Assemble paired-end reads
• Execute a shell script to automate a process
• Describe input and output files for amplicon workflows and scripts
• Describe the structure and components of a good mapping/metadata file
• Move sequences from compute resources to local computer
• Obtain summary information about sequence files (fasta, fna, fastq)
• Define operational taxaonomic units (OTUs)
• Align sequences, assign taxonomy, and build a tree with representative sequences from OTU definitions
• Calculate similarity of two samples (similarity matrices)
• Visualize comparative diversity across a priori categorical groups
• Convert .biom formatted OTU tables to text files for use outside of QIIME

Shotgun metagenomics & metatranscriptomics¶

• To assemble or not?
• To gene call or not?
• Kmers (“reference-free”)
• Taxonomic profiling
• Coverage estimation
• Reference mapping / alignment
• Gene Catalogues
• Marker-Based approaches
• Read-based approaches
• Functional profiling/annotation
• Functional hierarchies / ontologies
• Depth comparisons
• Binning
• Whole genome assembly and evaluation
• Pathway Analysis / Metabolic modeling
• Cross assembly

Links to resources that cover several for the above topics¶

• John Parkinson
• Reference Tutorial
• BioMAS
• Morgan Langille metagenomic tutorial
• Metagenomics Crash Course
• CABBIO
• TGAC

Learning objectives¶

Shotgun metagenomics & metatranscriptomics (Intro):

• Discern the difference between genetic potential and expressed genes.
• Evaluate the advantages and disadvantages between the metaT and metaG
• Aware of the cost and depth of sequencing differences for metaT and metaG

Reference mapping / alignment:

• Understand approaches to the elimination of host-associated or contaminating material
• Apply reference mapping to assess community structure
• Interpret the mapping outcomes as representative abundance of taxa.
• Compare the mapping approach with the “reference-free”.

Taxonomic profiling:

• Compare taxonomic profiling using metagenomics and amplicon-based approaches.
• Utilise both marked based and all-read methods for determining taxonomic profiles from metagenomic data
• Apply one or more methods to create a taxonomic profile
• Distinguish reference taxonomies and database limitations.
• Apply appropriate statistical methods for comparative analysis.
• Demonstrate knowledge of the limitations of quantifying microbes as relative abundances

Functional profiling:

• Recognise the role and application of different reference databases in functional assignments
• Use tools to perform functional annotations of nucleic acid and protein sequences found in metagenomics datasets.
• Demonstrate an understanding of the relationship between ontologies (e.g. GO) and functional assignments as a means of hierarchically viewing the data.
• Critically assess the functional assignments in terms of confidence, with clear understanding about the limitations of the algorithms and databases used.
• Apply pathway gap-filling methods to overcome low coverage sequencing

Links to resources that cover several for the above topics¶

• STAMP
• Dutilh
• Luis Pedro
• Stuart Jones EDAMAME
• BPA CSIRO

Learning objectives¶

Normalization / appropriate comparisons:

• basic: applying DeSEQ for normalization
• more advanced: raise awareness and discuss the possible limitations of DeSEQ and rarefaction for metagenomics analysis

Ecological measurements/indexes:

Alpha diversity (Chao, Simpson, etc. )

• basic: knowing that alpha diversity refers to diversity of a single sample. Understanding what the inputs are, that it can be computed/reasoned at different taxonomic levels (species, genus, …). Ability use some software package to compute.
• More advanced: Interpreting the values, knowing which measures are more robust to low abundance taxa, different sampling depths, &c. Understanding the effects of “dark matter” when using reference based methods (ie, microbes that do not match the reference being employed would not be counted towards diversity).
• Advanced: knowing the formulas and understanding the underlying ecological/mathematical assumptions.

beta diversity (bray curtis distances, Unifrac etc,)

• Basic: knowing that beta diversity refers to diversity across samples (differing definitions in the field). SImilar to alpha diversity, can be computed at different taxonomic (or functional) levels. Ability to use a software package.
• Intermediate: Understanding the effects of different normalization procedures.

Multivariate statistics (Differential expression, comparison)

• general statistics knowledge, unspecific to metagenomics (p-values, multiple comparison correction, difference between statistical significance and effect size, training/testing split in machine learning).
• Knowing how auto-correlation can lead to spurious correlations (ecological studies).
• More advanced: knowing which methods attempt to correct for auto-correlation and how to apply them in software.

Machine Learning

• Basic: knowing that an ordination plot represents distance between samples in a low dimensional space. Knowing that the output represents the input distance matrix (i.e., different input matrices would lead to different results: see discussion on beta diversity). Being able to generate a plot in software.
• Intermediate: knowing what the “fraction of explained variance” represents. Advanced: understanding the difference between the different methods.