Antisense Peptide Technology for Diagnostic Tests and Bioengineering Research

Antisense peptide technology (APT) is based on a useful heuristic algorithm for rational peptide design. It was deduced from empirical observations that peptides consisting of complementary (sense and antisense) amino acids interact with higher probability and affinity than the randomly selected ones. This phenomenon is closely related to the structure of the standard genetic code table, and at the same time, is unrelated to the direction of its codon sequence translation. The concept of complementary peptide interaction is discussed, and its possible applications to diagnostic tests and bioengineering research are summarized. Problems and difficulties that may arise using APT are discussed, and possible solutions are proposed. The methodology was tested on the example of SARS-CoV-2. It is shown that the CABS-dock server accurately predicts the binding of antisense peptides to the SARS-CoV-2 receptor binding domain without requiring predefinition of the binding site. It is concluded that the benefits of APT outweigh the costs of random peptide screening and could lead to considerable savings in time and resources, especially if combined with other computational and immunochemical methods.


Antisense Peptide Technology (APT)
The antisense peptide binding preference for the complementary sense sequence provides the opportunity to build a technology platform for the development and implementation of new immunochemical procedures and assays which use antisense peptides instead of the primary and/or secondary antibodies. APT is based on a heuristic algorithm for rational peptide design of the interacting ligand-receptor (acceptor) sequences specified by the complementary codons (Tables 1 and 2, Figure 1) . Heuristic methods reduce solution space by focusing on results based on the reduced set of criteria-in this case, complementarity rules defined by the standard genetic code (SGC) table [19,24].
Four main problems of sense and antisense peptide applications in immunochemistry are comparable to the reasons for the lack of success of the synthetic peptide vaccines [25], while the fifth topic addressed is related to the application of antisense/complementary sequences in bioengineering [18][19][20]24]: 1. reliance on continuous epitopes, 2. overconfidence in ligand specificity, 3. amino acid bias in characterizing ligand-acceptor (receptor) interactions, 4. difficulties in the estimation of structure-function relationships between specific ligand-acceptor (receptor) pairs, 5. amino acid coding, complementarity, and frameshifts.

Antisense Peptide Technology (APT)
The antisense peptide binding preference for the complementary sense sequence provides the opportunity to build a technology platform for the development and implementation of new immunochemical procedures and assays which use antisense peptides instead of the primary and/or secondary antibodies. APT is based on a heuristic algorithm for rational peptide design of the interacting ligand-receptor (acceptor) sequences specified by the complementary codons (Tables 1 and 2, Figure 1) . Heuristic methods reduce solution space by focusing on results based on the reduced set of criteria-in this case, complementarity rules defined by the standard genetic code (SGC) table [19,24].
Four main problems of sense and antisense peptide applications in immunochemistry are comparable to the reasons for the lack of success of the synthetic peptide vaccines [25], while the fifth topic addressed is related to the application of antisense/complementary sequences in bioengineering [18][19][20]24]: 1.
difficulties in the estimation of structure-function relationships between specific ligand-acceptor (receptor) pairs, 5.
Each of these specific problems is worth addressing.

Reliance on Continuous Epitopes
Epitopes and paratopes are not structural features of molecules [25,26]. They are entities characterized by a recognizable identity and defined by mutual complementarity [25,26]. Edmundson et al. [25][26][27] proposed the contact model of "flexible keys and adjustable locks" for epitope and paratope interaction. An epitope may be characterized as continuous and discrete [25,26]. Modeling of continuous epitopes by means of APT is often used for the sequences between 5 and 15 amino acids. The application of APT to discontinuous epitopes, between 10 and ≈20 amino acids, is more complicated, and similarly to Abs, often requires complex procedures that involve precise definition in structural terms, i.e., X-ray crystallography [7,20,25]. Discontinuous epitopes consist of (a) (b)

Amino Acid Bias in Characterizing Ligand-Acceptor (Receptor) Interactions
The huge difference in the number of possible antisense peptides available for sequence selection and possible database screening depends on the direction of the mRNA translation. The standard genetic code table specifies the translation of antisense (or complementary) peptides in two directions. Table 2 shows that 27 antisense amino acid pairs are derived by the 3′ → 5′ translation direction and significantly more (52 pairs) are obtained using the 5′ → 3′ direction algorithm [16][17][18][19][20]. The latter result is due to the fact that 5′ → 3′ antisense translation of the genetic code is based on 16 groups of codons, while 3′ → 5′ antisense translation depends on only four codon groups [19]. According to Siemion et al. [13,19], there are three main hypotheses concerning the interaction of sense-antisense peptides based on complementary coding principles.
The Mekler-Blalock antisense hypothesis is based on the hydropathic complementarity principle of sense and antisense peptide interactions, named Molecular Recognition Theory (MRT), which is independent of the direction of triplet reading, since the central (second) base of the coding triplet specifies the hydropathy of the amino acid [7][8][9]13,19].

Amino Acid Bias in Characterizing Ligand-Acceptor (Receptor) Interactions
The huge difference in the number of possible antisense peptides available for sequence selection and possible database screening depends on the direction of the mRNA translation. The standard genetic code table specifies the translation of antisense (or complementary) peptides in two directions. Table 2 shows that 27 antisense amino acid pairs are derived by the 3 → 5 translation direction and significantly more (52 pairs) are obtained using the 5 → 3 direction algorithm [16][17][18][19][20]. The latter result is due to the fact that 5 → 3 antisense translation of the genetic code is based on 16 groups of codons, while 3 → 5 antisense translation depends on only four codon groups [19]. According to Siemion et al. [13,19], there are three main hypotheses concerning the interaction of sense-antisense peptides based on complementary coding principles.
The Mekler-Blalock antisense hypothesis is based on the hydropathic complementarity principle of sense and antisense peptide interactions, named Molecular Recognition Theory (MRT), which is independent of the direction of triplet reading, since the central (second) base of the coding triplet specifies the hydropathy of the amino acid [7][8][9]13,19].
The Siemion hypothesis of sense-antisense peptide interaction is based on the periodicity of the genetic code, i.e., the Siemion one-step mutation ring of the code, and the resulting sense-antisense amino acid pairs are in most cases similar to the 3 → 5 translation direction [13,19].
The clustering of amino acid pairs, according to interaction preference, is defined by the complementary U ↔ A and C ↔ G bases of the second codon base. The second codon base, according to Woese, specifies the physicochemical properties of the amino acids [24,34]. Therefore, it is not surprising that diverse amino acid properties-like hydrophobicity, hydrophilicity, lipophilicity, and molecular descriptors of contact potential (Miyazawa-Jernigan), hydrophobic moment, and intrinsic disorder-follow the identical sense and antisense complementarity clustering scheme that is associated with molecular interaction at the peptide level (≥4 aa) [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24]. In a recent article, Štambuk et al. [19] emphasized that "the natural genetic coding algorithm for sense and antisense peptide interactions combines elements of amino acid physico-chemical properties, stereochemical interactions, and bidirectional transcription". The relationship of the genetic code and amino acid polarity with respect to protein structure and temperature conditions are discussed in reference [24] and the related Data in Brief article.

Difficulties in the Estimation of Structure-Function Relationships between Specific Ligand-Acceptor (Receptor) Pairs
The effects of an antisense ligand on its sense receptor (acceptor) may arise from the biological modulation and/or neutralization of the sense peptide effects by means of [6,7,17,35]: • peptides binding into molecular complexes (leaving none or low levels of sense peptide to elicit its own biological effects), • total or partial antagonization of the sense peptide receptor by means of its complexation with an antisense ligand, • combination of the first two factors, • other biological or biochemical effects of an antisense peptide that cannot be explained by the involvement of a sense peptide and its receptors (e.g., generation of bioactive antibodies to peptides and/or their complexes, cellular receptor, and growth factor modulation).
Each of those points should be carefully analyzed in the context of receptor binding and the biological effects observed under specific experimental designs. For example, in proliferative and biochemical studies involving cellular receptors and serum/plasma proteins, experimental results may be modified by the anticoagulant used-heparin, citrate, or EDTA [36,37].
A recent article by Bartonek et al. [43] showed that a frameshifting mechanism could be an effective evolutionary strategy for generating novel proteins with mostly unchanged physicochemical properties. Nevertheless, an important aspect of frameshift coding related to antisense/complementary sequences needs to be addressed. In 1996, Arques and Michel [41,42] identified a complementary circular code of trinucleotides (X) which on average has the highest occurrence in the reading frame (X 0 ) compared to the two shifted frames (X 1 and X 2 ).
This code was found in the protein coding genes of bacteria, archaea, eukaryotes, plasmids, and viruses [42,51]. It enables the reading frames to be retrieved in genes without start codons and with a window length of ≥13 nucleotides [41,42]. The frame X 0 consists of 12 amino acids (A, N, D, Q, E, G, I, L, F, T, Y, V), while frames X 1 (A, R, C, I, L, K, M, P, S, T, V) and X 2 (A, R, C, Q, G, H, L, P, S, W, Y) have 11 amino acids each [41,42,51]. With respect to the antisense codon and amino acid translation in the 5 → 3 direction, the X 0 frame of the circular code is self-complementary, and X 1 and X 2 frames are mutually complementary [41,42]. In 1999, Štambuk showed that the combinatorial necklace model enables the use of coding theory arithmetic in the analyses of the genetic code and circular code antisense translations [24,44,45,52].
Two seemingly opposite biological coding rules are characteristic for the interpretation of the SGC frameshifts and related mathematics-including complementary transformations within frames. They both deal with the mechanisms of translation error-control and flexibility and could have an important impact on SGC repertoire manipulations.
The first coding rule is that X 0 , X 1, and X 2 frames of the circular code distinguish three possible reading frames of the protein-coding sequence since hidden stop codons in X 1 and X 2 prevent off-X 0 -frame protein translation-this procedure is often named ambush hypothesis [53,54], and it is thought to ensure accurate translation.
Paradoxically, the second coding rule-related to SGC flexibility-is that stop codon readthrough may be promoted by the nucleotide environment, with glutamine (Q), tyrosine (Y), and lysine (K) inserted at UAA and UAG stop codons, whereas tryptophan (W), cysteine (C), and arginine (R) could be inserted at a UGA stop codon [55,56].
Considering bioengineering modeling, a reduced number of amino acids in frames X 0 , X 1 , and X 2 match the criteria for the use of simplified amino acid alphabets for engineering purposes and related sample space reductions [57]. Consequently, we measured the relationships of the main amino acid (aa) properties addressed by Bartonek et al. [43] in the frames X 0 , X 1 , and X 2 of the complementary circular code [41,42]. The factors of amino acid polarity, secondary structure, molecular volume, diversity, and electrostatic charge by Atchley et al. [58] were correlated to scales of nucleobase/amino acid interaction preferences for guanine (GUA), purines (PUR), and pyrimidines (PYR) [43,59].
A significant rise in the correlation of amino acid polarity to preference scales for guanine GUA, PUR, and PYR was observed in frame X 0 (Table 3). In frame X 1 (shifts +1 and −2), we found a strong correlation between codon and amino acid diversity factor and GUA, PUR, and PYR scales (Table 3). This observation is not surprising, since Atchley et al. [58] reported that diversity factor exhibits a highly significant correlation to amino acid physiochemical attributes and substitution matrices, and the X 1 frame is specified by the second codon base, which is associated with the majority of such information [24,34,60]. Table 3. Correlations between amino acid factors and preference scales in frames 0 and +1 (−2).

Parameter
Polarity ( However, in frame X 2 (shifts +2 and −1), correlations between physiochemical factors and nucleobase preference scales were not significant. This observation is in agreement with recent findings that, contrary to X 1 , the frame X 2 of the complementary circular code is less optimized than the SGC to reduce the effects of +2 and −1 frameshifts, in particular with respect to the physicochemical properties of amino acids [51].
A rise in correlation among amino acid factors and nucleobase preference scales in frames X 0 and X 1 of the circular codes may reflect the importance of the first two bases for the variables encoding scheme [24,34,60], and points to a possible application of GUA, PUR, and PYR scales [43,59] to different genetic code analyses. In our opinion, comparative investigations of complementary circular code and SGC-concerning frameshifts, errorcorrection, evolution, and biological engineering-seem to be justified.
As emphasized by Choi et al. [61], "ribosome is intrinsically susceptible to frameshift before its translocation and this transient state is prolonged by the presence of a precisely positioned downstream mRNA structure". Additionally, according to Rozov et al. [62], ribosome also "prohibits the G-U wobble geometry at the first position of the codon-anticodon helix". Therefore, it is not surprising that programmed ribosomal frameshifting enables reverse-genetics approaches and the construction of modified viruses with engineered deletions and/or foreign inserts [63].
Such engineering procedures could be used: 1. for artificial control of gene expression at the translation level, and 2. to generate differentiable marker vaccines and modified live virus vaccines [61,63]. More details on the challenges and perspectives of reverse vaccinology (RV) approaches may be found in Van Regenmortel [64] and Moxon, Reche, and Rappuoli [65].

An Example of SARS-CoV-2
One potential application of antisense peptide technology is related to the SARS-CoV-2 receptor-binding domain (RBD). The SARS-CoV-2 entry into human cells is characterized by the binding between viral spike (S) protein RBD and its receptor protein, i.e., angiotensinconverting enzyme II (ACE2) [66][67][68][69]. Consequently, a promising area of SARS-CoV-2 research is the disruption of SARS-CoV-2 binding to ACE2 by means of designer drugs and/or peptides [70][71][72]. Glasgow et al. [73] applied a stepwise engineering approach and generated enzymatically inactivated ACE2 variants, i.e., receptor traps, to potentially block RBD-ACE2 binding and entry into host cells to prevent infection. They used a position distant from the enzyme active site but situated at (or near) the place of enzyme contact with the virus-binding domain [73].
A similar approach may be taken by using an antisense peptide design. The ACE system was one of the classic targets for the investigation of sense-antisense peptide interactions decades before the COVID-19 crisis [3,6,[15][16][17]74], and Huang et al. [75] used APT translation in the 5 → 3 direction to screen a potential inhibitor for SARS-CoV.
Huang et al. [75] identified a dodecapeptide KKKKYRNIRRPG with high-binding affinity and specificity to the S protein of SARS-CoV and concluded that it could be used as a lead compound for further inhibitor studies. We additionally analyzed the docking of KKKKYRNIRRPG to the S1 protein of SARS-CoV-2 by means of a CABS-dock server ( Figure 3, Table A1, Table A2 and Table S1). CABS-dock is a global docking procedure with explicit fully flexible docking simulation and clustering-based scoring, in which receptor flexibility is limited by default to small backbone fluctuations (but can be increased to include selected receptor fragments) [28]. We observed docking of the peptide ligands to the 190 residue fragments (aa 334-523) of S1 RBD known to be responsible for virus binding (Table A1) [66,68,69]. The results presented in Figure 3 and Table A2 indicate that KKKKYRNIRRPG binds RBD S1 of SARS-CoV-2 at positions Y156 (aa 489), L159 (aa 492), Q160 (aa 493), and S161 (aa 494)-that belong to the region relevant for receptor binding [66,68,69,73]. This confirms the validity of the APT approach and applicability of the methods of Huang et al. [75] for SARS-CoV-2 molecular design.  (Table S1: model 1) [28,75]. Pairs of peptide/receptor residuals shown in magenta or blue (3′ → 5′ antisense); (b) Contact map of the interface between peptide and receptor residues in a complex closer than 4.5 Å (Tables A2 and S1: model 1).
APT, as a heuristic method, may also contribute to the rationalization of peptide library screening [18][19][20]24,32]. The heuristic approach to problem-solving is a practical method that employs an empirical procedure that may not always be optimal or perfect but is sufficient for finding a satisfactory approximate solution [76][77][78]. In computer science, bioinformatics, and mathematical optimization, this technique is often used as a shortcut for problem-solving in situations when classic methods are too slow or when they fail to provide the exact solutions [76][77][78]. The genetic coding algorithm for complementary (sense and antisense) peptide interactions [19] belongs to this class of so-called "heuristic algorithms". Their basic property is a trade-off between speed and completeness [78]. They are often used to solve NP-complete problems, i.e., decision problems without a known way to find a solution that is quick and efficient. The solution obtained  (Table  S1: model 1) [28,75]. Pairs of peptide/receptor residuals shown in magenta or blue (3 → 5 antisense); (b) Contact map of the interface between peptide and receptor residues in a complex closer than 4.5 Å (Table A2 and Table S1: model 1).
The validity of the 3 → 5 antisense approach was tested using the same methodology. We designed a 3 → 5 antisense paratope motif KLTIKGDVSIPKVGWLPQPIV complementary to the S1 RBD sequence of SARS-CoV-2 FNCYFPLQSYGFQPTNGVGYQ at sequence positions spanning from F153 to Q173 of the RBD S1, i.e., its aa 486-506 region ( Figure A1). The docking was done using a CABS-dock server ( Figure 4, Table A3). The results, presented in Figure 4 and Table A3, accurately predict the binding of 3 → 5 antisense peptide paratope KLTIKGDVSIPKVGWLPQPIV to its sense S1 RBD receptor epitope FNCYFPLQSYGFQPTNGVGYQ. Important contact positions are Y156 (aa 489), L159 (aa 492), Q160 (aa 493), S161 (aa 494), G163 (aa 496), and Y172 (aa 505), which are within the region that is relevant for receptor binding [66,68,69,73]-which points to the applicability of APT for the modeling of sense-antisense peptide interactions in SARS-CoV-2 research.

Peptide Modeling and Peptide-Protein Docking
The fragment of S1 RBD sequence aa 334-523 (190 aa, Table A1)-reported to be responsible for SARS-CoV-2 binding [66,68,69]-was used for protein modeling, design,  (Table S2: model 1) [28]. Pairs of peptide/receptor residuals: in 3 → 5 direction marked blue, in 5 → 3 direction marked red, otherwise magenta; (b) Contact map of the interface between antisense peptide and its receptor residues in a complex closer than 4.5 Å (Table A3 and Table S2: model 1). APT, as a heuristic method, may also contribute to the rationalization of peptide library screening [18][19][20]24,32]. The heuristic approach to problem-solving is a practical method that employs an empirical procedure that may not always be optimal or perfect but is sufficient for finding a satisfactory approximate solution [76][77][78]. In computer science, bioinformatics, and mathematical optimization, this technique is often used as a shortcut for problem-solving in situations when classic methods are too slow or when they fail to provide the exact solutions [76][77][78]. The genetic coding algorithm for complementary (sense and antisense) peptide interactions [19] belongs to this class of so-called "heuristic algorithms". Their basic property is a trade-off between speed and completeness [78]. They are often used to solve NP-complete problems, i.e., decision problems without a known way to find a solution that is quick and efficient. The solution obtained via a heuristic algorithm can be used directly, or it may represent a good point for the application of exact optimization algorithms. For example, to find all possible random amino acid binders for a peptide 5 aa in length, we need to examine 3.2 × 10 6 solutions (20 5 ). On the other hand, using APT, the range of possible solutions is 1-243 for the 3 → 5 and 1-1024 for the 5 → 3 translation, with ≈32 (≈2 5 ) as the average number of solutions. This represents a reduction of solution space by a factor of 1 × 10 5 . For longer sequences, this difference is even more pronounced. A heuristic solution could be easily checked experimentally, and APT is a typical example of that since it has been confirmed to be valid for >50 ligand-acceptor systems [4][5][6][7][8][9][12][13][14][15][16][17][18][19][20][21][22][23]. Computational models have their cost (approximately 10 USD/model) [79], which means that the reduction of possible solution space has a significant influence on the search outcome.
Two important S1 epitopes were determined with Expasy Server-ProtScale [81,82] using the hydrophobicity scale of Kyte & Doolittle, with a sliding block of 11 amino acids (Figure 6a,b). The probable binding site in the region 153-173 was confirmed by the Informational Spectrum Method (ISM, Figure 6c) using electron-ion interaction pseudopotential (EIIP) [18,83,84]. The ISM located the bioactive hot spot at the exposed (E) position Q165 (Q498) within the region of S1 RBD fragment 153-173, which corresponds to the S1 amino acid motif 486-506, empirically shown to be important for the virus-receptor binding [66,68,69,73]. Identification of B cell epitopes by means of the Bepi Pred Server [85], presented in Figure 6d, confirmed the importance of this region and the result of ISM.
The docking of the 5 → 3 antisense peptide KKKKYRNIRRPG, 3 → 5 antisense peptide KLTIKGDVSIPKVGWLPQPIV, and peptide LVMGLNVWLRYSK to the RBD of S1 SARS-CoV-2 protein was done using the CABS-dock server (Figures 3-5 and Tables A2-A4) [28]. CABSdock is a useful tool in the exploration of possible peptide binding sites [28,86]. It is a global docking procedure characterized by explicit, fully flexible docking simulation and clusteringbased scoring [28,86,87]. The CABS model is based on a knowledge-based parameterization of the molecular interactions, which consist of several statistical potentials [28,86]. It describes the excluded volumes of the united atoms, a model of the main chain hydrogen bonds, and the side-chain contact potentials [86]. The results of the CABS-dock procedure accurately predicted the binding of both antisense peptides and peptide LVMGLNVWLRYSK to the S1 RBD receptor epitope FNCYFPLQSYGFQPTNGVGYQ (Figures 3-5 and Tables A2-A4). (Figure 6a,b). The probable binding site in the region 153-173 was confirmed by the Informational Spectrum Method (ISM, Figure 6c) using electron-ion interaction pseudopotential (EIIP) [18,83,84]. The ISM located the bioactive hot spot at the exposed (E) position Q165 (Q498) within the region of S1 RBD fragment 153-173, which corresponds to the S1 amino acid motif 486-506, empirically shown to be important for the virus-receptor binding [66,68,69,73]. Identification of B cell epitopes by means of the Bepi Pred Server [85], presented in Figure 6d, confirmed the importance of this region and the result of ISM. The docking of the 5′ → 3′ antisense peptide KKKKYRNIRRPG, 3′ → 5′ antisense peptide KLTIKGDVSIPKVGWLPQPIV, and peptide LVMGLNVWLRYSK to the RBD of S1 SARS-CoV-2 protein was done using the CABS-dock server (Figures 3-5

Heuristic Antinsense Peptide Design (HAPD)
A simple heuristic antisense peptide design (HAPD) was completed using the optimization procedure based on three steps for complementary, i.e., sense-antisense, pairs selection (Scheme 1). The steps were denoted by colors (e.g., yellow, green, and turquoise). The first step, yellow, was used to model the antisense peptide "consensus skeleton", based on the selection of pairs specified by both complementary translation directions (3 → 5 and 5 → 3 , Table 2). The first step reduced the number of random peptides in the HAPD model by the factor of 4.096 × 10 15 , i.e., from 20 12 to 1. The second optimization step was green, and it specified the direction of translation for the "skeleton" (3 → 5 or 5 → 3 ). For the design of the antisense peptide KLTIKGDVSIPKVGWLPQPIV, we chose the 3 → 5 direction. The green criterion was the maximal hydropathy distance between polar-nonpolar amino acid clusters specified by the second base of the SGC table (A and U) and minimal hydropathy distance between neutral-neutral amino acid clusters specified by the second base of the SGC table (G and C) [7,8,17,19,24,88]. We applied the standard hydropathy distance criterion based on the Kyte and Doolittle scale [7,17,82]. For the clusters specified by the 3 → 5 complements of the modeled S1 sequence FNCYFPLQSYGFQPTNGVGYQ, the result is clear-QV, SS, and TW pairs were used ( Table 2). The second (green) step enabled further reduction of elements by the factor of 6.4 × 10 7 , i.e., from 20 6 to 1. In the third, turquoise, step there was only one possible antisense pairing left, containing the letters N, D, or E at the peptide aa position No. 7. This last step further reduced the number of random peptide pairs from 8000 (20 3 ) to only 3. Three amino acids (N, D, and E) have an identical hydropathy value (−3.5) and hydropathy difference (∆ = 7.3) in relation to their only matching 3 → 5 complementary transcript L (3.8) [7,82]. Consequently, we randomly selected aspartic acid (D) for the final antisense sequence KLTIKGDVSIPKVG-WLPQPIV. A similar strategy could also be applied to the 5 → 3 direction complements or to combinations of both procedures. For specific applications in immunochemistry and other disciplines, similar heuristic procedures for APD may also be combined with other heuristic bioinformatic methods such as, e.g., the basic local alignment search tool (BLAST) [18,20]. This procedure could also be used to model different natural or artificial (de novo) gene and protein constructs.
directions (3′ → 5′ and 5′ → 3′, Table 2). The first step reduced the number of rand peptides in the HAPD model by the factor of 4.096 × 10 15 , i.e., from 20 12 to 1. The sec optimization step was green, and it specified the direction of translation for the "skelet (3′ → 5′ or 5′ → 3′). For the design of the antisense peptide KLTIKGDVSIPKVGWLPQP we chose the 3′ → 5′ direction. The green criterion was the maximal hydropathy dista between polar-nonpolar amino acid clusters specified by the second base of the SGC ta (A and U) and minimal hydropathy distance between neutral-neutral amino acid clus specified by the second base of the SGC table (G and C) [7,8,17,19,24,88]. We applied standard hydropathy distance criterion based on the Kyte and Doolittle scale [7,17,82]. the clusters specified by the 3′ → 5′ complements of the modeled S1 seque FNCYFPLQSYGFQPTNGVGYQ, the result is clear-QV, SS, and TW pairs were u ( Table 2). The second (green) step enabled further reduction of elements by the facto 6.4 × 10 7 , i.e., from 20 6 to 1. In the third, turquoise, step there was only one poss antisense pairing left, containing the letters N, D, or E at the peptide aa position No This last step further reduced the number of random peptide pairs from 8000 (20 3 ) to o 3. Three amino acids (N, D, and E) have an identical hydropathy value (−3.5) hydropathy difference (Δ = 7.3) in relation to their only matching 3′ → 5′ complemen transcript L (3.8) [7,82]. Consequently, we randomly selected aspartic acid (D) for the f antisense sequence KLTIKGDVSIPKVGWLPQPIV. A similar strategy could also applied to the 5′ → 3′ direction complements or to combinations of both procedures. specific applications in immunochemistry and other disciplines, similar heur procedures for APD may also be combined with other heuristic bioinformatic meth such as, e.g., the basic local alignment search tool (BLAST) [18,20]. This procedure co also be used to model different natural or artificial (de novo) gene and protein constru Scheme 1. Heuristic method based on a three-step optimization procedure for APD.
A useful heuristic algorithm reduces the size of the solution space for a problem also enriches the density of hits in the subset. Instead of 20 n randomly generated pept for a sequence of length n within a full sample space, HAPD generates the average su number of 1.35 n peptides in 3′ → 5′ and 2.6 n peptides in 5′ → 3′ translation direction. C sequently, the probability of deriving a pentapeptide binder using the random pep design approach should be 3.125 × 10 −7 . In contrast to the random library approach, Scheme 1. Heuristic method based on a three-step optimization procedure for APD.
A useful heuristic algorithm reduces the size of the solution space for a problem and also enriches the density of hits in the subset. Instead of 20 n randomly generated peptides for a sequence of length n within a full sample space, HAPD generates the average subset number of 1.35 n peptides in 3 → 5 and 2.6 n peptides in 5 → 3 translation direction. Consequently, the probability of deriving a pentapeptide binder using the random peptide design approach should be 3.125 × 10 −7 . In contrast to the random library approach, the average probabilities to detect pentapeptide binders using HAPD 3 → 5 and 5 → 3 methods would be 0.223 and 0.0084, respectively. Therefore, it is not surprising that the validity of HAPD was confirmed for more than 50 ligand-acceptor systems .

Conclusions
The applicability of APT was confirmed recently for the magnetic particle enzyme immunoassay (MPEIA, Figure 2b) and immunohistochemical procedures [19,20]. This opens a perspective for the development of a new class of efficient immunochemical assays based on short peptide technology [18][19][20]. Additionally, it was also shown that modern computational methods enable a new approach to the studies of sense and antisense peptide interactions [20]. Several free web-based services for protein structure prediction and modeling (e.g., I-TASSER, Phyre2, PEP-FOLD 3, CABS-dock) enable accurate proteinpeptide docking, i.e., in silico search for the peptide binding sites [20,28,80,86,87,[89][90][91].
Small molecules and peptides may be also used for blocking protein-protein and protein-peptide interactions. In addition to NMR and X-ray crystallographic methods and mutational data, computational and virtual spectroscopy methods-such as the informational spectrum method (ISM)-could be also used to define hot spots in proteins [18,83,84]. An APT-based approach is also useful for peptide interaction and pharmacophore modeling [32,35]. The application of artificial proteins in the context of APT is also a plausible method to derive new antisense modulators of the protein interactions [19,24,88,91,92].
APT could be easily adapted to magnetic and polystyrene bead assays, conventional ELISAs, and multiplex assays, so it is possible to achieve two major lines of quick and sensitive assay development: 1. MPEIAs read with appropriate absorbance readers, and 2. Multiplex ELISAs read with appropriate imagers (e.g., with a high-resolution chemiluminescence readers for printed microtiter plates) [19,20,93,94].
Developing new immunoassays is important for situations such as the SARS-CoV-2 infection outbreak (and COVID-19 disease) due to the possibility to design-in a relatively short time-quick, inexpensive, and simple assays that could be automated to obtain medium/high throughput screenings of particular binders, peptide motifs, and antibodies, etc. If carefully selected, such laboratory techniques enable the experimental application of different laboratory procedures which, depending on the experimental design, may be used for: • selection of different targets and evaluation of complementary (sense-antisense) peptide binding; • quantification of specific antibodies, peptides, and proteins; • design of MPEIAs and Multiplex ELISAs tailored for a specific purpose.
The benefits of APT outweigh the costs of medium/high throughput screening and random peptide libraries and could lead to considerable savings in time and money. Practical applications and benefits of APT application are:

1.
Quick design and validation of the complementary ligands and acceptors; 2.
Computational validation and virtual screening of different protein and peptide structures; 3.
The tests can be produced in a short period of time; 5.
The tests will be made composite (according to the LEGO principle) and will consist of less expensive and commercially available components; 6.
The time required to obtain results is shorter (since no antibody production is needed); 7.
The test enables large quantity sample testing using standard laboratory equipment (since it does not require special reagents or complicated sampling processing); 8.
The tests are likely to prove important for the investigation of the immune response, disease pathogenesis, and clinical outcome of different infections; 9.
Designed antisense peptides (and anti-antisenses [21]) may also provide a basis for further development of vaccines and lead compounds for different diseases; 10. Detection of mutant strains is quicker since new antisense peptide motifs could be synthesized, evaluated for binding, and easily linked to magnetic particles in a short period of time, which avoids the antibody production process; 11. A green chemistry approach significantly reduces or avoids the loss of animal life.
Author Contributions: All authors listed have contributed substantially to the work reported. N.Š., P.K. and J.P. conceived and designed the study. N.Š. and P.K. performed molecular modeling. J.P. contributed to the production of graphics, as well as literature search and editing. All authors have read and agreed to the published version of the manuscript.  A   Table A1. Fragment of S1 RBD sequence aa 334-523 of SARS-CoV-2was used for CABS-dock analyses [28]. Table A2. Pairs of peptide/receptor residuals closer than 4.5 Å in the selected complex using the CABS-dock method [28]. Peptide sequence was KKKKYRNIRRPG [75] and receptor region was aa 334-523 of RBD S1 SARS-CoV-2.  Table A3. Sense-antisense pairs of peptide/receptor residuals closer than 4.5 Å in the selected complex using CABS-dock method [28]. Antisense peptide sequence was KLTIKGDVSIPKVGWLPQPIV, and the receptor sense region was aa 334-523 of RBD S1 SARS-CoV-2.  5  Table A4. Pairs of peptide/receptor residuals closer than 4.5 Å in the selected complex using CABSdock method [28]. Peptide sequence was LVMGLNVWLRYSK [29] and receptor was region aa 334-523 of RBD S1 SARS-CoV-2. (a) (b) Figure A1. (a) Receptor binding domain fragment aa 486-506 of S1 SARS-CoV-2 protein (PDB file: Table S5), and its antisense peptide translated in 3′ → 5′ direction (PDB file: Table S6) (b). Peptide structure prediction and modeling was done using I-TASSER (Iterative Threading ASSEmbly Refinement) [80].