Vaccines were used extensively before the antibiotics became accessible. Vaccination proves to be the most successful available strategy of an integrated prevention/therapeutic toolkit. It has significantly reduced the prevalence of a variety of infectious diseases such as bacterial and viral infections. It has slowed down the rate of development of resistant strains thereby preventing the further spread of several devastating infections globally (Andre et al., 2008; Rana and Akhter, 2016). A vaccine represents a biological formulation which upon administration to a given population can generate life time’s immunity against a particular disease (Mohan et al., 2013). First generation vaccines were developed using attenuated or inactivated strains of microbial pathogens. These have been reported as efficient for inciting both humoral and cellular immune responses (Seder and Hill, 2000). The second generation vaccine is composed of pathogen-derived purified components (devoid of the factors responsible for infection) instead of the whole microbial cells. These have been developed using novel recombinant proteins and DNA molecules (rDNA technology) as well as non-virulent but immunoprotective forms of microbial pathogens. The high-throughput sequencing and availability of complete genomic information have paved the way to a new ‘third generation’ of the vaccines (Seib et al., 2009). On vaccine administration, the vaccinated individual’s immune system encounters antigens expressed by disease-causing foreign pathogen and remembers it in form of immunological memory. This immunological memory, when encounters the real microbe expressing those antigens, there is production and activation of highly specific memory T lymphocytes, B lymphocytes and natural killer cells (Ratajczak et al., 2018). This rapidly generates an effective immune response against the microbial pathogen (Ottenhoff and Kaufmann, 2012). Hence, the most important job of vaccines is to expose the vaccinated individuals with much milder and non-virulent pathogenic antigens to generate immunological memory without actually causing the disease. A brief history of major breakthroughs in vaccine development has been illustrated in Figure 1.

The most commonly used first generation vaccine against the mycobacterial pathogens is Bacillus Calmette-Guerin (BCG). It is composed of attenuated (non-virulent) strains of M. bovis. In the following subsections, we are summarizing the current use and protection status provided by the BCG vaccine.

Genomics may be described as a comprehensive analysis of an organism’s complete genome. The genome represents the complete set of DNA/genes (coding and non-coding) present in a cell or organism. There are approximately 3.2 billion bases and an estimated 20000 protein coding genes in humans. Traditionally, genes were analyzed individually but with the advent of microarray technology, genome-wide differential expression studies are made possible in recent years. DNA microarrays measure the subtle differences among DNA sequences (genetic variations) like small-scale insertion/deletions, polymorphic repetitive elements, SNPs and microsatellite variation among different individuals. The most common type of genetic variation is single nucleotide polymorphisms (SNPs). SNP occurs when one nucleotide in the genome is substituted for another and differs between members of the same species (Horgan and Kenny, 2011). This change results in an alternative codon and hence different amino acid which may be of particular interest when associated with complex mycobacterial diseases (Stucki and Gagneux, 2012). Various abnormalities like chromosomal insertions or deletions can be identified with more advanced microarray based comparative genomic hybridization (aCGH). CGH is a popular molecular cytogenetic technique for genome-wide screening of cells for chromosomal copy number variations. It uses two differentially labeled genomic DNAs: test and control sample which are simultaneously cohybridized to metaphase chromosomes. The differentially colored fluorescent signal intensity of the fluorophore labeled test DNA relative to control sample DNA is linearly plotted along the length of each chromosome to provide a cytogenetic representation of copy number variation between the two sources (Kallioniemi et al., 1992). However, CGH shows a very limited resolution of alterations of approximately 5–10 Mb (Kirchhoff et al., 1998; Lichter et al., 2000). To overcome this limitation, a more advanced high-resolution platform is known as array CGH (aCGH) has been developed. Instead of targeting metaphase chromosomes, it utilizes cloned DNA elements (known as probes) arrayed on a slide as the targets for analysis (Lucito et al., 2003). These probes are from different origins and vary in size like oligonucleotides (25–85 base pairs), bacterial artificial chromosomes (BACs; 80,000–200,000 base pairs). The probes used in aCGH are far smaller than the metaphase chromosomes which allows greater mapping resolution in aCGH than the traditional CGH. The mapping resolution depends upon both the probe size and genomic distance between DNA probes (Theisen, 2008).

The human genome project initiated in 1990 annotated the DNA sequence of the complete euchromatic human genome. Since then, the sequencing technologies [Sanger and next-generation sequencing (NGS)] have remained the hottest topic in the field of genomics research (Gasperskaja and Kućinskas, 2017). In the modern DNA sequencing era, with the ongoing technological advancement in the field of genomics, the sequencing technologies are revolutionizing the genome research especially with the high-throughput NGS (HT-NGS). It has a wide range of applications such as: chromatin immunoprecipitation (‘ChIP’) with DNA microarray (‘chip’) also known as ‘ChIP-on-chip’ and ChIp-sequencing (ChIP-seq) (Pareek et al., 2011).

Historically in 1975, the “first generation” DNA sequencing technique, known as ‘Sanger’s method’ or ‘dideoxy chain termination method,’ was developed based on specifically labeled chain terminating dideoxynucleotides (ddNTPs) incorporated by DNA polymerase during in vitro DNA synthesis. The fundamental principle behind this targeted sequencing technique is that the ddNTPs are different from dNTPs at 3′ carbon and fail to make phosphodiester bond with the next nucleotide which terminates the nucleotide chain elongation and hence replication halts. In this way, different bands of varying lengths are generated which are then separated on a polyacrylamide gel. After band separation, a laser reads the gel to detect the fluorescent intensity of each band in the form of colored peaks in a chromatogram. These colored peaks represent the nucleotide in that specific location in the DNA sequence (Russell, 2002).

Although Sanger method has proven useful in performing a thorough analysis of DNA, its use has been limited because of the high cost and size limitation. The Sanger method can only read short pieces of DNA (1000–1200 base pair) and the quality degrades after 700–900 base pairs. More recently, to overcome major stumbling blocks of first generation sequencing, new generations of sequencing techniques have been introduced which include NGS. NGS is capable of sequencing millions of DNA fragments through a massively parallel analysis with much reduced cost producing huge sequencing data. It has proven to be the new game changer for DNA sequencing. Although NGS exploits the principle similar to that of Sanger’s method of sequencing, which relies on the separation of labeled DNA elements by electrophoresis and identification of emitted signals, NGS uses array-based sequencing. It combines Sanger’s techniques (sequencing, separation and detection) for analysis of millions of samples in parallel at reduced cost with high throughput. It involves three steps: library preparation- small fragments of DNA created using random fragmentation (enzymatically or sonification) and ligated with custom linkers, amplification- done by PCR (emulsion PCR or bridge PCR), sequencing- DNA sequenced using “sequencing by synthesis” or “sequencing by ligation” (Zhang et al., 2011; Ari and Arikan, 2016). The ever growing field of sequencing has sparked an enormous range of applications of NGS technology in different research fields such as elucidation of the molecular basis of genetic diseases, infectious diseases and cancer (Del Vecchio et al., 2017).

ChIP assays are the most invaluable methods to identify the protein binding sites on DNA. ChIp-seq couples ChIP assays with NGS to investigate the genome-wide DNA binding sites for physical binding interactions of transcription factors. In ChIP-seq, formaldehyde fixation is used to irreversibly cross-link proteins to their bound DNA. The cross-linked chromatin is sheared with sonication or restriction enzymes to generate small fragments of DNA associated with a particular protein of interest followed by immunoprecipitation with desired antibody-bound magnetic beads. For NGS library preparation, the precipitated genomic DNA is used as input and is sequenced for DNA binding site analysis (Gasperskaja and Kućinskas, 2017). A more recent approach named ‘ChIP-on-chip’ combines ChIP with microarray analysis. In this method, the precipitated DNA fragments are hybridized to a microarray chip for analysis. It generates a global genome-wide chromatin maps depicting genome-wide binding sites of protein which may help to identify the functional elements in the complete genome. While this technique proved to be a revolutionary approach to study large genomic regions, it suffered from certain technical limitations such as high cost and requirement of a large amount of DNA thus extensive amplification leading to biasness and allelic variants hindered by cross-hybridization (Mikkelsen et al., 2007).

Hence, genomic analysis techniques provide an enormous amount of valuable information which may be translated in form of novel biomarkers to expedite antigen discovery. The genomic analysis usually begins with the identification and selection of potential coding regions. Along with this, attribution of functions to the selected novel proteins on the basis of sequence homology followed by a reverse genetic evaluation to characterize the complete repertoire of unannotated hypothetical proteins may be carried out (Geluk et al., 2014). Among the major mycobacterial infections, the complete genome sequence of Mtb H37Rv (Krogh et al., 2001) and CDC-1551 strains (Betts, 2002) and M. bovis AF2122/97 strain (Garnier et al., 2003) have revolutionized a big impact on the pace of anti-mycobacterial drug discovery. The genome sequence of M. leprae strain TN (Singh and Cole, 2011) has also been established. Using various in silico approaches, the whole set of protective antigens can easily be identified from the microbial pathogen’s genome without even cultivating it in the laboratory (Rappuoli, 2000). Hence, genome analysis can circumvent the laborious, costly and time-consuming conventional approaches and may pave the way to a better and faster discovery of antigenic targets against mycobacterial infections.

The transcriptome reflects the set of all RNA molecules or transcripts in a cell or organism. Transcriptomics aim to study all species of transcripts including mRNAs, non-coding RNAs and small RNAs produced in a cell of an organism at a specific time (Wang et al., 2009; Kunnath-Velayudhan et al., 2017; Lowe et al., 2017). Transcriptomics analysis has played a central role in unraveling the gene expression during a particular physiological condition and deciphering the intricacies of regulations at the transcriptional level. Expression profiling of transcripts could be targeted to explore the specific genes which show expression or overexpression in host and pathogens simultaneously representing a complete atlas of hot-spots of host-pathogen interactions (Kaiser et al., 2004). Several technologies in the field of transcriptomics have emerged to derive and quantify the RNA content, including hybridization-based and sequence-based approaches. The dominant contemporary techniques like microarrays typically measure the transcripts by hybridization of fluorescently labeled cDNA against a custom-made array of complementary probes or high-density spotted oligonucleotide microarrays. The transcriptional profiling by hybridization-based approaches is labor saving with high throughput and reduced cost. However, these suffer from some limitations such as they can detect only known sequences, high background levels generally lead to cross-hybridization and interfere with detection. Although microarray technology continues to support transcriptomics research, the advent of sequence based approaches have dramatically expanded transcriptomics in the past few years (Wang et al., 2009).

In contrast to classical hybridization techniques, the high-tech sequencing based approaches directly determine the nucleic acid sequence of cDNA. In earlier times, Sanger’s method was used to sequence cDNA or EST libraries (Gerhard et al., 2004), but this method was expensive with relatively low throughput. To overcome this, high throughput tag-based transcriptome profiling methods were developed which included cap analysis gene expression (CAGE), serial analysis gene expression (SAGE), and massively parallel signature sequencing (MPSS). Since, these methods were based on conventional Sanger sequencing technique, these were expensive and failed to map some of the short tags to the reference genome. Additionally, they failed to analyze transcript isoforms which are generally indistinct from each other. These limitations reduced the potential use of conventional sequencing technology as transcriptome profiling method (Wang et al., 2009).

Recently, the newly developed high-throughput DNA sequencing techniques have enabled highly sensitive analysis for mapping, profiling and quantifying RNAs. This rapidly growing transcriptome profiling technique is known as RNA-Seq or whole transcriptome shotgun sequencing (WTSS). RNA-Seq utilizes an NGS platform and is replacing gene expression microarrays at a high rate. For this method, RNA (fractionated or total) is first converted to cDNA molecules with the help of reverse transcriptase followed by PCR amplification. Each molecule is then sequenced using NGS sequencing platform. Following sequencing, a genome-scale transcription map is generated when the output reads are aligned to reference transcripts or reference genome (Wang et al., 2009). RNA-Seq is an effective and excellent approach for transcriptome profiling of host and pathogen simultaneously. Moreover, this technique has also been successfully used to compare HCV- or HIV-infected T-cells to uninfected T-cells in vitro. It has revealed differentially expressed transcripts of the virus and the metabolic effects of viral infection on the target cells (Lefebvre et al., 2011).

Exploiting the above mentioned transcriptomic techniques, a number of studies have been reported describing the identification of various RNA molecules involved in different regulatory networks responsible for the virulence of pathogenic mycobacterial species. RNA-Seq and high-density tiling arrays have deciphered a large repertoire of previously unknown non-coding mycobacterial RNA including novel antisense transcripts, 5’ and 3’ untranslated regions and intergenic small RNAs (sRNAs) (Arnvig and Young, 2012; Michaux et al., 2014).

Non-coding RNA (ncRNA) molecules represent RNA transcripts that are generally not translated into a protein. Although, exceptionally, some ncRNA may contain an ORF and may translate into a polypeptide chain. There are different classes of ncRNA defined on the basis of cellular processes such as ncRNAs involved in mRNA translation (rRNAs and tRNAs), splicing (small nuclear RNAs -snRNAs), modification of rRNAs (small nucleolar RNAs-snoRNAs) and gene expression regulation (microRNAs-miRNAs, piwi-interacting RNAs-piRNAs, long non-coding RNAs-lncRNAs (Arnvig and Young, 2012; Qureshi and Mehler, 2012).

The sRNAs are generally the non-coding small transcripts in the range of 50–250 nucleotides in length. They are involved in gene silencing and post-transcriptional regulation and are generally encoded opposite the ORF (cis-encoded) or between ORF (trans-encoded) (Haning et al., 2014). The first mycobacterial stress regulatory sRNA was identified in 2009. The cDNA libraries of low molecular weight Mtb transcriptomes (exponential and stationary phase) were analyzed to identify 5 trans-encoded and 4 cis-encoded sRNAs in Mtb H37Rv (Arnvig and Young, 2009). Until now, a total of nearly 200 sRNAs have been identified in Mtb (Gerrick et al., 2018). The sRNAs discovered so far have gained significant attention, especially in pathogens as regulators of transcription factors, pathogenic genes, outer membrane adaptation to stress conditions like the variation in environmental pH, temperature and anaerobic stress (Haning et al., 2014; Michaux et al., 2014).

miRNAs are evolutionarily conserved small non-coding RNA molecules of 20–24 nucleotide length. These have been reported to play a regulatory role at the post-transcriptional level by binding to the 3’-UTR of their target mRNAs and inhibiting their translation. In pathogenic mycobacterial species, these miRNAs have been demonstrated to play an important role as immunomodulators by regulating the genes expressed by immune cells of the host and in-turn supporting its growth and survival inside the host. In recent studies, it has been shown that the innate immune response generated against TB is regulated by these miRNAs. Additionally, miRNAs differential expression during TB reflects disease progression and are capable of distinguishing active TB from latent TB (Palazzo and Lee, 2015; Ahluwalia et al., 2017; Sabir et al., 2018).

Hence, the uniquely expressed RNAs identified by high-throughput transcriptomic methods provide new insights into pathogenesis and could be targeted as potential biomarkers or as therapeutic agents against mycobacterial diseases.

Proteome reflects the entire set of expressed proteins in a cell, tissue or organism at any given time (Theodorescu and Mischak, 2007). Proteomics covers a number of different aspects of protein function, including structural proteomics: large-scale analysis of protein structures, expression proteomics: large-scale analysis of protein expression and interaction proteomics: large-scale analysis of protein interactions. The main aim of proteomics is to study and characterize the information flowing within a cell or organism in the form of protein pathways and networks, (Petricoin et al., 2002) in order to understand the functional importance of proteins (Vlahou and Fountoulakis, 2005). Proteomics studies provide a deep understanding of the various virulent factors in different disease causing microorganisms and can aid the discovery of suitable markers as novel therapeutic agents (Fournier and Raoult, 2011).

Conventionally, different chromatographic methods have been used for purification and separation of proteins such as gel filtration/size exclusion chromatography (SEC), ion exchange chromatography (IEC) and affinity chromatography (Jungbauer and Hahn, 2009; Voedisch and Thie, 2010; Hage et al., 2012). To analyze selective proteins, techniques like western blotting and ELISA have been widely used. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), two-dimensional gel electrophoresis (2-DE) and two-dimensional differential gel electrophoresis (2D-DIGE) techniques have also been used to separate complex protein samples (Marouga et al., 2005; Issaq and Veenstra, 2008). An emerging proteomics technique, named as protein microarrays or protein chips provides a versatile platform to analyze proteins on large scale. While mass spectrometry, another analytical technique, is used to analyze complex protein mixtures on the basis of the mass-to-charge ratio of charged particles with high sensitivity (Yates, 2011). Additionally, Edman degradation is used to sequence amino acids in a particular protein (Smith, 2001). To quantify global changes in protein numbers, a number of peptide quantitation techniques have been developed including, metabolic based labeling [stable isotope labeling with amino acids in cell culture (SILAC)] and isotope-coded affinity tag (ICAT) labeling, isobaric mass tagging [isobaric tag for relative and absolute quantitation (iTRAQ)], chemical and enzymatic derivatization [quantitation by isobaric terminal labeling (QIRT)] (Ong and Mann, 2006; Shiio and Aebersold, 2006; Wiese et al., 2007; Kroksveen et al., 2015) etc. The three-dimensional structures of proteins are obtained using two popular experimental high-throughput techniques: nuclear magnetic resonance (NMR) spectroscopy and X-ray crystallography (Smyth and Martin, 2000; Aslam et al., 2017).

With the advent of proteomics techniques, their applications have been wide-ranging and expanded in almost every discipline of biological sciences. In silico analysis of the available proteomic data has defined several new ‘omes’ having potential antigenic targets. These include the exportome (Van Ooij et al., 2008), surfome (Sargeant et al., 2006), and interactome (Sanchez et al., 1999). The surfome or surface proteome of several pathogens has been identified using proteolytic shaving (Rodríguez-Ortega et al., 2006) and biotinylation (Cullen et al., 2005). Currently available proteomic techniques exploiting peptide libraries and antibody microarrays have been used to analyze Mtb proteome to identify potential antigen candidates (Kunnath-Velayudhan and Porcelli, 2013). There was a report where workers have annotated most potential subunit vaccine candidates by comparing the mycobacterial proteomes of Mtb and M. bovis BCG. They observed that Rv3407, a DNA vaccine candidate could be used to improve the overall efficacy of the existing BCG vaccine (Mollenkopf et al., 2004). Others have also discovered novel antigenic markers from the identified secreted and transmembrane proteins employing proteomics approach- glutathione S-transferase (GST) fusion protein purification strategy (Zhou et al., 2015). Similarly, Mtb Rv0444c, Rv3692, and Rv2031c have been identified as possible candidate biomarkers from an analysis performed through MALDI-TOF-MS (Zhang et al., 2012). These may be targeted for the development of diagnostic assays against TB in the near future.

In the present “omics” era, metabolomics is rapidly emerging as a field of science to study the systematic identification, quantification and analysis of cellular metabolites within a given biological system (cell, tissue, organ, biological fluid or organism) at any given time. It is a collection of sophisticated analytical techniques to study the outcome of complex networks of biochemical reactions providing an understanding of the cellular physiology on a global biochemical scale (Mirsaeidi et al., 2015; Nandakumar et al., 2015).

Some of the modern analytical platforms used to study metabolite profiles include proton nuclear magnetic resonance (1H-NMR) spectroscopy, gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectroscopy (LC-MS). These have been used to provide sensitive and reliable detection of metabolites to be exploited in diagnosis and prognosis of several infectious diseases (Weiner et al., 2012; Ghannoum et al., 2013; Mickiewicz et al., 2014). The metabolomics studies of mycobacterial pathogens are still in their nascent period of development. The recent studies about Mtb metabolome have provided unique insights into the biochemical composition, organization, activity and regulation of its physiological network (Nandakumar et al., 2015). The metabolites arising from a mycobacterial pathogen or its host have yielded important information describing undefined metabolism and pathogenic characteristics linked to the pathophysiology of mycobacterial infections (Miyamoto et al., 2016). du Preez and Loots have analyzed the sputum of 34 TB patients with 2D-gas chromatography time-of-flight mass spectrometry (GC-MS) (du Preez and Loots, 2013). They successfully identified 22 metabolites (14 Mtb metabolites and 8 host-related metabolites) as potential biomarkers against TB (du Preez and Loots, 2013). Similarly, in another study, using the same analytical tool, it was reported that 2-acetylamino-2-deoxy-b-D-glucopyranose, a-L-mannopyranose and D-galactose-6-deoxy could be targeted to differentiate TB infected patients from non-infected persons (Cha et al., 2009; O’Sullivan et al., 2012). In a different liquid chromatography-mass spectrometry (LC-MS) based metabolomics study, it was observed that rpoB mutations change the Mtb metabolic profile and it plays an important role in its metabolism. A total of 99 molecular features were found different in the Mtb rifampin-resistant strains (Bisson et al., 2012). In a different study, non-targeted ultrahigh-pressure liquid chromatography time-of-flight mass spectroscopy (UPLC-TOF-MS) was exploited to distinguish a cohort of patients infected with leprosy having bacterial index < 1 from those with a bacterial index > 4 (increased metabolites: polyunsaturated fatty acids, eicosapentaenoic acid and docosahexaenoic acid) (Al-Mubarak et al., 2011).

Compared to the other ‘omics’ technologies, metabolomics has fewer limitations and offers potential advantages in terms of specificity and sensitivity (van Ravenzwaay et al., 2007). As metabolomics captures the snapshot of the metabolic status of the genes providing useful insights about the biochemical networks under study, it allows more complete understanding of cell functions perhaps far more than genomics, transcriptomics or proteomics can (Lindon et al., 2003).

Today, with the advent of genomic technology, the genome-based antigen selection is possible and allows the discovery of antigen and vaccine design. One approach that mines pathogenic bacterial genomes for antigen discovery is known as “Reverse Vaccinology” (RV). RV has emerged as an effective strategy that uses bioinformatics techniques with the aim to identify highly protective and immunogenic peptides encoded by immunologically exposed pathogenicity factors by screening the entire genomes of microbial pathogens (Movahedi and Hampson, 2008; Seib et al., 2012; Donati and Rappuoli, 2013; Figure 2). RV based antigen discovery pipeline involves genome sequence analysis for the identification of antigenic proteins (surface exposed or secreted) expressed by the pathogen, their cloning and expression followed by synthetically producing each protein. The best selected candidates could be tested in the clinical trials for validating their immunogenicity after in vitro immunogenicity examination in cells and animal models. The identified antigens may be targeted for vaccine discovery. To date, RV has been targeted to devise universal and effective vaccines against bacterial pathogens for which the discovery of vaccines was previously impossible. Among these, N. meningitidis serogroup B (MenB) (Pizza et al., 2000), against which there was no effective vaccine, was the first pathogen targeted for the development RV based human vaccine (Delany et al., 2013; Rappuoli et al., 2016).

With the help of RV, whole-genome studies are now being more focused on the development of target specific epitope-based vaccines. An epitope or an antigenic determinant is the specific part of antigen interacting with the immune system (T-cell, B-cell and antibodies). The antigenic epitopes elicit an immune response by interacting with the CD8+ T immune cells and CD4+ T immune cells and may be used in ‘reverse’ to target novel antigens (Sette and Rappuoli, 2010). The immune cells- B and T lymphocyte play a major role in antigen recognition and elicitation of the immune system. The B lymphocytes are the plasma cells that produce antibodies when a foreign antigen triggers immune system and function as humoral immunity component of the adaptive immune system. The epitopes of the antigens are identified by the paratopes of antibody. The T lymphocytes play a central role in cell-mediated immunity. Hence, the prediction of the immunodominant T and B cell epitopes plays an important role in the determination of the peptide-based candidate vaccines (Kanampalliwar et al., 2013).

Based on RV, a number of web-based programs have also been developed to assist the scientific community in identifying potential vaccine candidates against mycobacterial infections. These include MycobacRv, Violin, VaxiJen, and MtbVeb, etc. MycobacRv is an RV based database of potential mycobacterial adhesins vaccine candidates from 23 strains and other species of mycobacteria. It houses detailed epitope information from the predicted adhesins and surface-localized/extracellular proteins which may further facilitate the development of epitope-based mycobacterial vaccines (Chaudhuri et al., 2014). Vaccine Investigation and Online Information Network (VIOLIN) is another web-based database that integrates vaccine literature mining, vaccine data curation and storage. It also provides an analytical platform for potential vaccine target prediction against various infectious agents (He et al., 2014). Likewise, VaxiJen is another useful resource available online for the prediction of protective antigens and subunit vaccines. The predictions are alignment independent and solely based on the physicochemical properties of the target proteins (Doytchinova and Flower, 2007). MtbVeb is a comprehensive database for designing novel vaccines against 59 existing and emerging Mtb strains employing antigen, strain and epitope based approaches (Dhanda et al., 2016). A growing number of studies reporting antigen identification published in the literature have provided valuable insights into RV based vaccine research. Some of them have been discussed in the coming sections of this review paper.

The available high-throughput ‘omics’ approaches have made it possible to identify potentially important biomarkers in various microbial pathogens in a much smaller time than the conventional approaches. The wide availability of data generated by these ‘omics’ technologies offer ample opportunities to unravel the disease mechanisms but also present the scientific community with significant challenges to extract the knowledge from such huge data and its application for the welfare of the society.

In genomics, the pathogenic microorganisms with larger genomes, that fails to be cultured in vitro or if there are no animal models available, may not be suitable for antigen discovery utilizing RV because of the huge number of possible targeted proteins with unknown function (Schussek et al., 2014). In the case of transcriptomics, the information generated from deep sequencing studies need in vivo validation and also require validation for multiple isolates of the microbial pathogen (Schussek et al., 2014).

Similarly, although proteomics offers advantages in antigen discovery, it still suffers from certain limitations. While performing proteomics analysis, the organism is allowed to grow in highly favorable conditions (in vitro) and is generally isolated at a specific phase of the cell cycle which certainly does not depict the in vivo environment of that organism (Singh et al., 2015). Furthermore, the proteomics studies may not be suitable enough to identify protein complexes which are resistant to proteases as reported earlier for pili associated proteins, which have been demonstrated as potential vaccine candidates for Staphylococcus aureus and Streptococcus pneumonia (Schussek et al., 2014). Moreover, the proteomics approach gives a limited level of understanding of the protein level events of microorganisms since the mRNA transcription of a gene necessarily does not give an estimation of its translated protein level. The reason could be: the transcribed mRNA might degrade quickly or it might get translated into protein ineffectively or alternative splicing might result in the generation of multiple proteins. Another reason could be the post-translational modifications of proteins which might result in an inactive protein (Kornblihtt et al., 2013). Another major limitation of the proteomics approach is many proteins are involved in complex formation to become completely functional (Srinivas et al., 2002). Additionally, the secondary and tertiary structures of proteins are often difficult to maintain during their analysis. These generally get denatured by the action of enzymes, heat or by external stress. The proteins of low abundance are often found difficult to detect as these cannot be amplified like DNA. Like in plasma, cytokines are present in very low quantity (1–5 pg/mL) and proteomics tools can analyze proteins mostly located at the higher end of the concentration spectrum. Hence, to study these low abundant proteins, the high abundant proteins are removed from plasma. However, this removal is often accompanied by the loss of several potentially important biomarkers resulting from co-removal of antigenically important proteins bound to the high-abundance proteins (Granger et al., 2005; Cho, 2007). For these reasons explained above, very often, the proteomics experiments performed in one laboratory are poorly reproducible in other laboratories.

Nevertheless, the metabolomics key features for several diseases (Monteiro et al., 2013; Aretz and Meierhofer, 2016) have been reported, the potential bottlenecks still exist at various levels of quality biomarker identification. It is hampered by the huge and dynamic variation in the metabolic levels between people, tissues and various time points. The other bio-molecular states like the genome, transcriptome and proteome are comparatively much more stable than the vastly fluctuating metabolites (Aretz and Meierhofer, 2016).

Hence, to fulfill the huge demand for novel robust biomarkers to curb the mycobacterial infections, different ‘omics’ platforms must together be integrated to reveal, assess and track down the novel molecular patterns reflecting the disease-perturbed networks.

Virulent factors represent the molecules essential for the growth of microbial pathogens which allow them to succeed and establish disease inside the host (Rana et al., 2015b). Earlier, the pathogenicity of bacteria was reported to be linked to toxins (Peterson, 1996) but later, it was considered to originate from the presence of various virulence determinants (Smith, 2003). Thus, it was concluded that targeting these potentially virulent factors would stop the disease establishment and would enable a rapid development of novel vaccines, antibiotics and new screening tests. The three main approaches that have been used for the identification of virulence genes from the complete genome involves: homology search with the experimentally characterized virulent factors (Rana et al., 2015c), identifying genes located in different pathogenic genomic islands (Akhter et al., 2007, 2008, 2012; Che et al., 2014) and the third approach involves identification of the virulence genes by genome comparison of strains having different pathogenicity profiles (virulent versus avirulent strains). Using an in silico approach, a set of 189 putative vaccine candidates have been identified from the complete Mtb genome (3989 gene products) (Zvi et al., 2008). A total of 40 promising therapeutic targets were identified in M. abscessus using novel hierarchical in silico approach and these may be exploited for novel drug discovery (Shanmugham and Pan, 2013). In an another in silico study performed on Mtb, 99 putative lipoproteins, playing important role in virulence, were identified using various bioinformatics utilities like TrEMBL database (Boeckmann et al., 2003), ScanProsite tool (Gattiker et al., 2002), SignalP (Nielsen et al., 1997), and TMHMM program (Krogh et al., 2001; Sutcliffe and Harrington, 2004). Similarly, in 2 different studies performed in silico, a total of 48 lipoproteins in Mtb, 25 lipoproteins in M. leprae, 75 lipoproteins in M. avium, 97 lipoproteins in M. marinum and 61 lipoproteins in M. smegmatis were computationally identified utilizing LipoP (Juncker et al., 2003; Rezwan et al., 2007). The pathogenic proteins targeting the host cell compartments like host mitochondria during infection are also the most commonly targeted virulent factors. A number of in silico proteome-wide studies have reported the potential mitochondria targeting proteins of the microbial pathogens (Moreno-Altamirano et al., 2012; Forrellad et al., 2013; Rana et al., 2015b). Forrellad et al., computationally identified 19 mitochondria targeting proteins from Mtb H37Rv virulent strain by utilizing the MitoProt program (mitochondria targeting proteins prediction) (Claros and Vincens, 1996), PSORT II prediction algorithm (sub-cellular localization) (Horton and Nakai, 1997) and SignalP (signal peptide sequence prediction) (Nielsen et al., 1997; Moreno-Altamirano et al., 2012). In a similar in silico approach, we have reported 46 MAP proteins as potential host mitochondria targeting proteins by employing different bioinformatics algorithms in tandem (Rana et al., 2015b). Firstly, complete MAP proteome was screened to detect the signal peptide sequence utilizing program SignalP and the identified exportome was analyzed for mitochondrial import signal screened through MitoProt II, TargetP and TPpred program (Savojardo et al., 2014). 46 MAP mitochondria targeting proteins were successfully identified. Out of these, 20 MAP proteins were defined as putative endotoxins from DBETH database (Chakraborty et al., 2011) and 14 MAP proteins as exotoxins by BTXpred tool (Saha and Raghava, 2007) which may be acting as potential virulent factors involved in MAP pathogenicity (Rana et al., 2015b).

A ‘secretome’ of an organism represents the total secretory proteins that are being released into the external milieu. This group of proteins is commonly known as excretory/secretory (ES) proteins and is important for the establishment of pathogenic infection within the host (Gomez et al., 2015; Rana et al., 2016). The SASPs include secretory proteins and surface-associated proteins like lipoproteins and OMPs. These SASPs are nowadays considered as promising targets for antigen discovery. These offer ample opportunities for the development of new therapeutic solutions against different clinical infections as the SASPs including ES proteins that are present at the interface of host-pathogen interaction and may also function as immune modulators of the host cells (Zagursky and Russell, 2001). They also help in the pathogen survival inside the host organism and act as virulence factors.

We have earlier reported novel and much advanced in silico approaches (Rana et al., 2014) for the proteome-wide identification of SASPs of MAP, M. leprae and Mtb (Rana et al., 2016). The approach exploits the cardinal sequence and structural features of SASPs from mycobacteria. The exportome of the MAP, M. leprae and Mtb was first identified employing Target P1.1 program followed by transmembrane helix prediction by TMHMM and HMMTOP program. The selected proteins were further analyzed for the presence of α helix and β sheet by utilizing the JPRED3 (Cole et al., 2008) program and amphiphilicity computation using Vogel and Jahnig algorithm (Vogel and Jähnig, 1986). Further, lipoproteins were predicted by PRED-LIPO (Juncker et al., 2003) program and sub-cellular localization of proteins was done using PSORTb followed by identification of non-classical secretory proteins employing SecretomeP program. The performed proteome-wide analysis identified 57 OMPs, 38 lipoproteins, 63 secretory proteins in the MAP; 19 OMPs, 17 lipoproteins, 11 secretory proteins in M. leprae; 36 OMPs, 47 lipoproteins and 49 secretory proteins in Mtb. Similar in silico studies have been conducted on various pathogenic genomes and proteomes to identify the repertoire of SASPs which represented novel candidates as virulence factors. These include: Taenia solium (Gomez et al., 2015), Phytophthora infestans (Raffaele et al., 2010), Yersinia pestis (Yen et al., 2007), Xanthomonas citri (Ferreira et al., 2016), Coxiella burnetii (Ferreira et al., 2016), and enteric pathogens including Shigella spp, E. coli, Vibrio cholerae, Yersinia enterocolitica (Hashmi et al., 2010), Salmonella spp., and Anaplasma marginale (Palmer et al., 2012).

Epitope mapping is one of the keystone steps to be considered while designing an effective potent vaccine (Palmer et al., 2012). It has remarkable advantages over the long established conventional methods since it is the most cost effective, highly specific and competent strategy to generate a specific desired long lasting immunity in the host. It also helps to avoid unwanted autoimmune responses. With the advent of diverse bioinformatics tools, epitopes are nowadays can easily be mapped from the whole genomes of microbial pathogens by performing in silico analysis, without immediate reference to the peptide fragments origin. Several immunoinformatics methods have been employed for designing a highly efficient vaccine that must be capable of generating a protective B and T-cell immune response (Davies and Flower, 2007; Rana et al., 2016). Numerous vaccine related studies integrated in silico RV approach to discover putative vaccine candidates against diverse pathogens.

In case of mycobacterial infections, RV studies reported that sxL, PE26, PPE65, PE_PGRS49, PBP1 and Erp were the six proteins identified with antigenic epitopes from Mtb, that could be targeted to design novel and more efficient vaccines against TB (Monterrubio-López, 2015). Eight proteins (MAP2698c, MAP2312c, MAP3651c, MAP2872c, MAP3523c, MAP0187c and the hypothetical proteins MAP3567 and MAP1168c) were also identified with highly immunogenic epitopes in the MAP as potential vaccine candidates for studying antibody and cell-mediated immune responses within infected hosts (Gurung et al., 2012). In our previous work, we have integrated biological knowledge together with bioinformatics tools to design a much more advanced methodology pipeline for epitope mapping of the MAP (Rana et al., 2015b) and M. leprae OMPs (Rana et al., 2016). Moreover, our earlier studies reported 83 potential OMPs from a total of 4356 MAP proteins, out of which 57 MAP proteins were identified as a core set of putative OMPs (Rana et al., 2014). The identified OMPs were first analyzed to identify the host homologous proteins and proteins with significant similarity to closely related Mycobacterium taxa for excluding them to prevent any potential cross-reactivity using BLAST analysis. Further, the non-homologous proteins were subjected to immunoinformatic analyses for the prediction of T-cell (MHC I: artificial neural network approach) (Nielsen et al., 2003; Tenzer et al., 2005); MHC II: consensus approach (Wang et al., 2008) and B-cell epitopes ElliPro suite (Ponomarenko et al., 2008). Similarly, RV has been successfully applied against various other pathogens for identification of suitable antigens for vaccine development such as Dichelobacter nodosus (Myers et al., 2007), Pasteurella multocida (Al-Hasani et al., 2007) and Mtb (Kundu et al., 2016).