1. Alignment-based approaches

• Phylogenomics is the study of evolutionary relationships via comparative analysis of genome-scale data, intersecting the fields of evolution and genomics.

• The first and key step for these studies is the comparative analysis of DNA and amino acid sequences. To achieve it, several programs were developed to perform sequence alignment.

• This method positions the biological sequences’ building blocks to identify regions of similarity that may have consequences for functional, structural, or evolutionary relationships.

1. Alignment-based approaches

Several alignment-based tools were created such as:

• Sequence similarity search tools (e.g. BLAST)

• Multiple sequence aligners (e.g. ClustalW, Muscle or MAFFT)

• Whole-genome aligners (e.g. progressive Mauve)

1.1 BLAST example

1. Alignment-based approaches

• Alignment-based programs assume that homologous sequences include a series of linearly arranged and more or less conserved sequence stretches (which is termed collinearity). For example, viral genomes exhibit great variation in the number of genetic elements.

• These approaches also depend on assumptions about the evolution of the sequences that are being compared. Small changes to input parameters can greatly affect the alignment.

• Finally, these approaches are memory and time consuming, where an accurate multiple sequence alignment cannot be solved in a realistic time frame.

2.1 Alignment-free approaches

• Any method of quantifying sequence similarity/dissimilarity that does not use or produce alignment at any step of algorithm application.

• Computationally less expensive;

• Resistant to shuffling and recombination events;

• Don't depend on assumptions regarding the evolutionary trajectories of sequence changes.

2.2 Categories of Alignment-free approaches

2.2.1 Word-based methods

2.2.2 Match length methods

How are AF being used for NGS analysis?

• Variant calling
  • Traditionally mapping-based detection (GATK HaplotypeCaller and Samtools mpileup);
  • AF methods allow for genotyping directly from NGS data, perfoming 1-2 orders or magnitude faster.

• Taxonomic profiling
  • Assignment of taxonomic labels wit near perfect accuracy (99%).
  • CLARK, Kraken and Mash.

• Assembly
  • Correction of sequencing errors in raw reads.

3. How well do AF methods work?

• Vinga and Almeida (2004) and Hohl et al. (2007) showed that AF may perform better than traditional methods in case of proteins that underwent domain shuffling events;

• Dai and colleagues (2008) demonstrated that AF methods detected statistically relevant similarities in sequence compositions in contrast to traditional methods that showed only limited correspondence recognizable by alignments;

• Bernard and colleagues (2016) showed that AF methods:

• Most sensitive to the extent of sequence divergence;

• Less sensitive to low and moderate frequencies of horizontal gene transfer;

• Most robust against genome rearrangements.

3. How well do AF methods work?