**Algorithm steps:**

Scanning method (SISCAN) [102], LARD [103], Topal/Difference of Sums of Squares

breakpoint positions using HMM and then identifies the recombinant sequence using various methods such as phylogenetic profiling (PHYLPRO) [105] and Visual

signals are then inferred. It involves sequential disassembly of the identified re‐ combinant sequences into respective components and iteratively rescanning the resulting expanded dataset until no further recombination signals are evident.

**ii.** Following the detection of a 'recombination signal', RDP4 determines approximate

**iii.** The minimum number of recombination events that are needed to account for these

RDP4 package provides a unified interface for multiple methods and facilitates visualization

**i.** The genomic dataset up to 200 million nucleotides can be analysed and is reported

**ii.** Recombination analysis is likely to fail in case of poor alignments, if recombinant

Natural selection is one of the fundamental evolutionary processes that shape the genetic structure of viral populations. The ratio of non-synonymous substitution rate (*dN*) to synon‐ ymous substitution rate (*dS*) is a useful means to infer selection pressure based on a codon alignment for a particular gene. Positive selection (*dN/dS* > 1) increases the frequency of advantageous alleles, whereas the negative selection (*dN/dS* < 1) is responsible for purging

Broadly, the selection pressure can be classified as pervasive and episodic. Pervasive selection acts across all the lineages in a phylogenetic tree, whereas the episodic selection operates on a few lineages of a tree. Various statistical methods for analysis of pervasive and episodic selection are available at the Datamonkey web-server of Hypothesis testing using Phylogenies

**Principle:** This method belongs to a class called counting methods [110]. It is suitable for pervasive selection analysis and involves estimating the number of non-synonymous and synonymous changes that have occurred at each codon throughout the evolutionary history of the sample. It involves reconstructing the ancestral sequences using likelihood-based

sequences are used as reference and sequences having ambiguous characters are

(DSS) [104] and DNA distance plot, are used.

188 Next Generation Sequencing - Advances, Applications and Challenges

Recombination Detection (VisRD) [106].

of recombination events using genomic data (up to 2,500 sequences).

to have operational limits for large genomic datasets.

**Salient feature:**

**Limitations:**

included.

(removal) of deleterious alleles.

(HyPhy) software package [107–109].

method [111].

*4.5.1. Single Likelihood Ancestor Counting (SLAC)*

**4.5. Methods for selection pressure analysis**


#### *4.5.2. Fixed-Effect Likelihood (FEL) and Internal Fixed-Effect Likelihood (IFEL)*

**Principle:** These belong to a class of methods called 'fixed effects'. It analyses pervasive selection and involves fitting substitution rates on a site-by-site basis by assuming that the synonymous substitution rate is the same for all sites. Thus, FEL and IFEL assume the same *dN/dS* (*ω*) ratio, which is applicable to all branches and to interior branches, respectively [111].

#### **Algorithm steps:**


FEL method: For every site, based on the parameter estimates obtained using nucleotide- and codon-fit procedure, two rate parameters namely *α* and *β* are first fitted independently and then under the constraint of *α* = *β*. Here, the parameter α represents the instantaneous synonymous site rate, while *β* represents the instantaneous non-synonymous site rate. Furthermore, LRT is performed to infer whether *α* is different from *β* and a *p*-value is com‐ puted. If the *p*-value is significant, the site is classified based on whether *α* > *β* (indicates negative selection) or *α* < *β* (indicates positive selection).

IFEL method: It differs from FEL in following aspects:

