Ivan Mazoni
Research Subjects
Ivan Mazoni has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Goran Neshich
Computational Biology Research Group, Embrapa Digital Agriculture, Av. André Tosello 209, Campinas, São Paulo 13083-886, Brazil
4 shared publications - José Augusto Salim
Plant Biology Department, Institute of Biology, State University of Campinas, São Paulo, Brazil
4 shared publications - Folorunsho Bright Omage
Computational Biology Research Group, Embrapa Digital Agriculture, Av. André Tosello 209, Campinas, São Paulo 13083-886, Brazil; Biological Chemistry Laboratory, Department of Organic Chemistry, Institute of Chemistry, University of Campinas (UNICAMP), R. Monteiro Lobato 270, Campinas, São Paulo 13083-970, Brazil
2 shared publications - Fábio Rogério de Moraes
Multiuser Center for Biomolecular Innovation, Department of Physics, Institute of Biosciences, Languages and Exact Sciences, Sao Paulo State University (UNESP), Sao Jose do Rio Preto, SP, Brazil
2 shared publications Goran Neshich
Embrapa Digital Agriculture
2 shared publicationsLuiz Borro
Claro beOn
2 shared publications
Latest Publications
- Open Access
- 15 January 2026
- Published by Oxford University Press (OUP) in Database: The Journal of Biological Databases and Curation
- Vol. 2026
STING-ExositeDB: An AI-assisted curated database of protein exosites for drug discovery
Motivation: Protein exosites are secondary binding sites that recruit macromolecular partners and control molecular recognition across protein families. Despite their therapeutic potential demonstrated by approved drugs such as bivalirudin, systematic characterization remains limited by fragmented literature and inconsistent reporting. Existing databases lack exosite-specific curation, residue-level contact mapping, and structure–function integration. Results: We present ExositeDB, a curated database of protein exosites built through AI-assisted curation combining large language model extraction with expert validation. Our three-pass pipeline acquires papers from multiple databases, extracts data using GPT-4o with confidence scoring, validates findings against PDB structures, and links all claims to source text. The current release contains 525 expert-validated exosite records spanning 280 unique proteins, each annotated with residue-level positions, partner types, and functional roles. Quality is maintained through multi-tier confidence scoring integrating experimental methodology (35% weight), structural validation (25%), literature consistency (20%), and terminology precision (20%). The database follows FAIR principles through structured metadata, versioned releases, persistent identifiers, and an open REST API. Structural coverage includes 395 unique PDB structures supporting structure–function validation. ExositeDB provides training data for AI-driven exosite prediction, supports rational design of selective modulators, and enables analysis of exosite-mediated regulatory networks.
- 15 January 2026
- Published by Oxford University Press (OUP) in Database: The Journal of Biological Databases and Curation
- Vol. 2026
Open AccessSTING-ExositeDB: An AI-assisted curated database of protein exosites for drug discovery
Motivation: Protein exosites are secondary binding sites that recruit macromolecular partners and control molecular recognition across protein families. Despite their therapeutic potential demonstrated by approved drugs such as bivalirudin, systematic characterization remains limited by fragmented literature and inconsistent reporting. Existing databases lack exosite-specific curation, residue-level contact mapping, and structure–function integration. Results: We present ExositeDB, a curated database of protein exosites built through AI-assisted curation combining large language model extraction with expert validation. Our three-pass pipeline acquires papers from multiple databases, extracts data using GPT-4o with confidence scoring, validates findings against PDB structures, and links all claims to source text. The current release contains 525 expert-validated exosite records spanning 280 unique proteins, each annotated with residue-level positions, partner types, and functional roles. Quality is maintained through multi-tier confidence scoring integrating experimental methodology (35% weight), structural validation (25%), literature consistency (20%), and terminology precision (20%). The database follows FAIR principles through structured metadata, versioned releases, persistent identifiers, and an open REST API. Structural coverage includes 395 unique PDB structures supporting structure–function validation. ExositeDB provides training data for AI-driven exosite prediction, supports rational design of selective modulators, and enables analysis of exosite-mediated regulatory networks.
- Open Access
- 2 July 2025
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 26 (4)
STINGAllo: a web server for high-throughput prediction of allosteric site-forming residues using internal protein nanoenvironment descriptors
Allosteric regulation is essential for modulating protein function and represents a promising target for therapeutic intervention, yet the complex dynamics of the protein nanoenvironment hinder the reliable identification of allosteric sites. Traditional pocket-based predictors miss $\sim $18% of experimentally confirmed sites that lie outside surface invaginations. To overcome this limitation, we developed STINGAllo, an interactive web server that introduces a residue-centric machine-learning model. Using 54 optimized internal protein nanoenvironment descriptors, STINGAllo predicts allosteric site-forming residues at single-residue resolution. By integrating hydrophobic interaction networks, local density, graph connectivity, and a unique “sponge effect” metric, STINGAllo detects allosteric sites independently of surface geometry, including concave pockets, flat surfaces, or even cryptic regions. It achieves a success rate of $\sim $78% on benchmark datasets, substantially outperforming existing methods with a 60.2% overall success rate compared with 21.1%–24.2% for contemporary pocket-based predictors. Our analysis further reveals that nearly 52.7% of unique proteins in the Protein Data Bank [(PDB); 119 851 entries, 14 November 2024] contain at least one chain with a predicted allosteric site. STINGAllo accepts protein structures via PDB identifiers or custom uploads, provides interactive 3D visualization of predicted pockets, and supports integration into computational pipelines through a RESTful application programming interface. Overall, STINGAllo bridges advanced computational prediction with user-friendly design, offering a robust tool expected to deepen understanding of protein regulation and accelerate allosteric drug discovery. The server is freely accessible at https://www.stingallo.cbi.cnptia.embrapa.br/.
- 2 July 2025
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 26 (4)
Open AccessSTINGAllo: a web server for high-throughput prediction of allosteric site-forming residues using internal protein nanoenvironment descriptors
Allosteric regulation is essential for modulating protein function and represents a promising target for therapeutic intervention, yet the complex dynamics of the protein nanoenvironment hinder the reliable identification of allosteric sites. Traditional pocket-based predictors miss $\sim $18% of experimentally confirmed sites that lie outside surface invaginations. To overcome this limitation, we developed STINGAllo, an interactive web server that introduces a residue-centric machine-learning model. Using 54 optimized internal protein nanoenvironment descriptors, STINGAllo predicts allosteric site-forming residues at single-residue resolution. By integrating hydrophobic interaction networks, local density, graph connectivity, and a unique “sponge effect” metric, STINGAllo detects allosteric sites independently of surface geometry, including concave pockets, flat surfaces, or even cryptic regions. It achieves a success rate of $\sim $78% on benchmark datasets, substantially outperforming existing methods with a 60.2% overall success rate compared with 21.1%–24.2% for contemporary pocket-based predictors. Our analysis further reveals that nearly 52.7% of unique proteins in the Protein Data Bank [(PDB); 119 851 entries, 14 November 2024] contain at least one chain with a predicted allosteric site. STINGAllo accepts protein structures via PDB identifiers or custom uploads, provides interactive 3D visualization of predicted pockets, and supports integration into computational pipelines through a RESTful application programming interface. Overall, STINGAllo bridges advanced computational prediction with user-friendly design, offering a robust tool expected to deepen understanding of protein regulation and accelerate allosteric drug discovery. The server is freely accessible at https://www.stingallo.cbi.cnptia.embrapa.br/.
- 1 June 2025
- Published by Bentham Science Publishers Ltd. in Current Nanomedicine
- Vol. 15 (3) , 267-285
Comprehensive Analysis of the Distinct Nano Environments Characteristics Containing the Different Secondary Structure Elements: α-Helices, β-Sheets, and Turns
Background: This work is the third part of our initiative to fully describe the internal protein nano environments (NEs) for the three existing types of secondary structure elements (SSE). In our previous work, the NE of both the α-helix and the β-sheet were analyzed. The focus of this and previous research is improving our understanding of the SSEs: α-helices, β-sheets and turns, within protein structures. We found that the structural similarities between turns and α-helices are very high and turns may be considered as the “incomplete” initiation of α-helices. The knowledge we were able to compile during this work might be essential for predicting a tertiary structure of proteins, with higher precision and subsequently being in a more favourable position with regard to designing drugs for certain protein structure/function related diseases. Considering that the formation of secondary structure elements is a crucial step in the general folding of protein 3D structure, an important contribution of this work is augmenting the efficiency of estimating if modelled structures, for example, are predicting SSEs in full agreement to necessary/required/ sufficient values of respective SSEs nanoenvironment descriptors. During exactly that modelling phase, preceding the more precise final 3D protein structure construction, our Dictionary of Most Relevant Nanoenvironment Descriptors is able to answer some fundamental questions regarding SSE correctness with regard to nanoenvironment characteristics found to be generally required. Expanding this vision, our current work is part of an effort by our laboratory to create a “dictionary of internal protein nanoenvironments” - DIPN. The ten most studied internal protein nanoenvironments are described in DIPN in physicochemical and structural terms, and this knowledge is now available to be used to aid drug design - probably the most important area of application for the results we are presenting here. Methods: In the current paper, STING´s database of physical-chemical and structural descriptors was used to gather the necessary information to characterize the NE of loops, or, as they are often called, turns. Given that approximately 20% of all protein-type residues form turns, research in this field is essential, and analysis of the obtained results will further contribute to our comprehension of how proteins fold. In addition, the results in this paper will contribute to the better training of algorithms that evaluate the degree of overall protein structure quality and, consequently, structure prediction. This is currently very important given we are witnessing a revolution in algorithms employing artificial intelligence for protein structure prediction. Powered by the STING’s database (wide-ranging protein structure information source), statistical testing was used to retrieve a set of descriptors that fully delineate the NE of turns. By collecting such data, it is then possible to list the variances with respect to the NE of α-helices and β-sheets and, by doing so, establish the most relevant NE descriptors (MRND) for each of the three SSEs. Results: The results show that the α-helical and β-sheet Nes, as well as the amino acid residue composition, all behave in a similar fashion as a “key and lock” system. In other words, it is necessary for a set of specific descriptors to assume respective specific values (within the bounds of a very definite value region) to construct the specific secondary structure element NE at a certain protein location. Conclusion: Consequently, there is a set of descriptors that act together and are required to satisfy specific conditions for secondary structure element occurrences. The very same requirement, we found, occurs in the case of turns.
- 1 June 2025
- Published by Bentham Science Publishers Ltd. in Current Nanomedicine
- Vol. 15 (3) , 267-285
Comprehensive Analysis of the Distinct Nano Environments Characteristics Containing the Different Secondary Structure Elements: α-Helices, β-Sheets, and Turns
Background: This work is the third part of our initiative to fully describe the internal protein nano environments (NEs) for the three existing types of secondary structure elements (SSE). In our previous work, the NE of both the α-helix and the β-sheet were analyzed. The focus of this and previous research is improving our understanding of the SSEs: α-helices, β-sheets and turns, within protein structures. We found that the structural similarities between turns and α-helices are very high and turns may be considered as the “incomplete” initiation of α-helices. The knowledge we were able to compile during this work might be essential for predicting a tertiary structure of proteins, with higher precision and subsequently being in a more favourable position with regard to designing drugs for certain protein structure/function related diseases. Considering that the formation of secondary structure elements is a crucial step in the general folding of protein 3D structure, an important contribution of this work is augmenting the efficiency of estimating if modelled structures, for example, are predicting SSEs in full agreement to necessary/required/ sufficient values of respective SSEs nanoenvironment descriptors. During exactly that modelling phase, preceding the more precise final 3D protein structure construction, our Dictionary of Most Relevant Nanoenvironment Descriptors is able to answer some fundamental questions regarding SSE correctness with regard to nanoenvironment characteristics found to be generally required. Expanding this vision, our current work is part of an effort by our laboratory to create a “dictionary of internal protein nanoenvironments” - DIPN. The ten most studied internal protein nanoenvironments are described in DIPN in physicochemical and structural terms, and this knowledge is now available to be used to aid drug design - probably the most important area of application for the results we are presenting here. Methods: In the current paper, STING´s database of physical-chemical and structural descriptors was used to gather the necessary information to characterize the NE of loops, or, as they are often called, turns. Given that approximately 20% of all protein-type residues form turns, research in this field is essential, and analysis of the obtained results will further contribute to our comprehension of how proteins fold. In addition, the results in this paper will contribute to the better training of algorithms that evaluate the degree of overall protein structure quality and, consequently, structure prediction. This is currently very important given we are witnessing a revolution in algorithms employing artificial intelligence for protein structure prediction. Powered by the STING’s database (wide-ranging protein structure information source), statistical testing was used to retrieve a set of descriptors that fully delineate the NE of turns. By collecting such data, it is then possible to list the variances with respect to the NE of α-helices and β-sheets and, by doing so, establish the most relevant NE descriptors (MRND) for each of the three SSEs. Results: The results show that the α-helical and β-sheet Nes, as well as the amino acid residue composition, all behave in a similar fashion as a “key and lock” system. In other words, it is necessary for a set of specific descriptors to assume respective specific values (within the bounds of a very definite value region) to construct the specific secondary structure element NE at a certain protein location. Conclusion: Consequently, there is a set of descriptors that act together and are required to satisfy specific conditions for secondary structure element occurrences. The very same requirement, we found, occurs in the case of turns.
- Open Access
- 29 March 2025
- Published by Oxford University Press (OUP) in NAR Genomics and Bioinformatics
- Vol. 7 (2) , lqaf054
Learning to utilize internal protein 3D nanoenvironment descriptors in predicting CRISPR–Cas9 off-target activity
Despite advances in determining the factors influencing cleavage activity of a CRISPR–Cas9 single guide RNA (sgRNA) at an (off-)target DNA sequence, a comprehensive assessment of pertinent physico-chemical/structural descriptors is missing. In particular, studies have not yet directly exploited the information-rich internal protein 3D nanoenvironment of the sgRNA–(off-)target strand DNA pair, which we obtain by harvesting 634 980 residue-level features for CRISPR–Cas9 complexes. As a proof-of-concept study, we simulated the internal protein 3D nanoenvironment for all experimentally available single-base protospacer-adjacent motif-distal mutations for a given sgRNA–target strand pair. By determining the most relevant residue-level features for CRISPR–Cas9 off-target cleavage activity, we developed STING_CRISPR, a machine learning model delivering accurate predictive performance of off-target cleavage activity for the type of single-base mutations considered in this study. By interpreting STING_CRISPR, we identified four important Cas9 residue spatial hotspots and associated structural/physico-chemical descriptor classes influencing CRISPR–Cas9 (off-)target cleavage activity for the sgRNA–target strand pairs covered in this study.
- 29 March 2025
- Published by Oxford University Press (OUP) in NAR Genomics and Bioinformatics
- Vol. 7 (2) , lqaf054
Open AccessLearning to utilize internal protein 3D nanoenvironment descriptors in predicting CRISPR–Cas9 off-target activity
Despite advances in determining the factors influencing cleavage activity of a CRISPR–Cas9 single guide RNA (sgRNA) at an (off-)target DNA sequence, a comprehensive assessment of pertinent physico-chemical/structural descriptors is missing. In particular, studies have not yet directly exploited the information-rich internal protein 3D nanoenvironment of the sgRNA–(off-)target strand DNA pair, which we obtain by harvesting 634 980 residue-level features for CRISPR–Cas9 complexes. As a proof-of-concept study, we simulated the internal protein 3D nanoenvironment for all experimentally available single-base protospacer-adjacent motif-distal mutations for a given sgRNA–target strand pair. By determining the most relevant residue-level features for CRISPR–Cas9 off-target cleavage activity, we developed STING_CRISPR, a machine learning model delivering accurate predictive performance of off-target cleavage activity for the type of single-base mutations considered in this study. By interpreting STING_CRISPR, we identified four important Cas9 residue spatial hotspots and associated structural/physico-chemical descriptor classes influencing CRISPR–Cas9 (off-)target cleavage activity for the sgRNA–target strand pairs covered in this study.
- Open Access
- 16 October 2023
- Published by Springer Nature in Research Square
Comprehensive analysis of the distinct nano environments characteristics containing the different secondary structure elements: α- helices, β-sheets, and turns
This work is the third part of our initiative to fully describe the internal protein nano environments (NEs) for the three existing types of secondary structure elements (SSE). In our previous work, the NE of both the α-helix and the β-sheet were analysed. The knowledge acquired in this research is important considering that secondary structure element formation is a crucial step in protein folding and an important phase that precedes the final 3D protein structure. In the current paper, STING´s database of physical-chemical and structural descriptors was used to gather the necessary information to characterize the NE of loops, or, as they are often called, turns. Given that approximately 20% of all protein-type residues form turns, research in this field is essential, and analysis of the obtained results will further contribute to our comprehension of how proteins fold. In addition, the results in this paper will contribute to the better training of algorithms that evaluate the degree of overall protein structure quality and, consequently, structure prediction. This is currently very important given we are witnessing a revolution in algorithms employing artificial intelligence for protein structure prediction. Powered by the STING’s database (wide-ranging protein structure information source), statistical testing was used to retrieve a set of descriptors that fully delineate the NE of turns. By collecting such data, it is then possible to list the variances with respect to the NE of α-helices and β-sheets and, by doing so, establish the most relevant NE descriptors (MRND) for each of the three SSEs. The results show that the α-helical and β-sheet Nes, as well as the amino acid residue composition, all behave in a similar fashion as a “key and lock” system. In other words, it is necessary for a set of specific descriptors to assume respective specific values (within the bounds of a very definite value region) to construct the specific secondary structure element NE at a certain protein location. Consequently, there is a set of descriptors that act together that are required to satisfy specific conditions for secondary structure element occurrences. The very same requirement, we found, occurs in the case of turns.
- 16 October 2023
- Published by Springer Nature in Research Square
Open AccessComprehensive analysis of the distinct nano environments characteristics containing the different secondary structure elements: α- helices, β-sheets, and turns
This work is the third part of our initiative to fully describe the internal protein nano environments (NEs) for the three existing types of secondary structure elements (SSE). In our previous work, the NE of both the α-helix and the β-sheet were analysed. The knowledge acquired in this research is important considering that secondary structure element formation is a crucial step in protein folding and an important phase that precedes the final 3D protein structure. In the current paper, STING´s database of physical-chemical and structural descriptors was used to gather the necessary information to characterize the NE of loops, or, as they are often called, turns. Given that approximately 20% of all protein-type residues form turns, research in this field is essential, and analysis of the obtained results will further contribute to our comprehension of how proteins fold. In addition, the results in this paper will contribute to the better training of algorithms that evaluate the degree of overall protein structure quality and, consequently, structure prediction. This is currently very important given we are witnessing a revolution in algorithms employing artificial intelligence for protein structure prediction. Powered by the STING’s database (wide-ranging protein structure information source), statistical testing was used to retrieve a set of descriptors that fully delineate the NE of turns. By collecting such data, it is then possible to list the variances with respect to the NE of α-helices and β-sheets and, by doing so, establish the most relevant NE descriptors (MRND) for each of the three SSEs. The results show that the α-helical and β-sheet Nes, as well as the amino acid residue composition, all behave in a similar fashion as a “key and lock” system. In other words, it is necessary for a set of specific descriptors to assume respective specific values (within the bounds of a very definite value region) to construct the specific secondary structure element NE at a certain protein location. Consequently, there is a set of descriptors that act together that are required to satisfy specific conditions for secondary structure element occurrences. The very same requirement, we found, occurs in the case of turns.
- Open Access
- 30 December 2020
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 15 (12) , e0244315
A comparison between internal protein nanoenvironments of α-helices and β-sheets
Secondary structure elements are generally found in almost all protein structures revealed so far. In general, there are more β-sheets than α helices found inside the protein structures. For example, considering the PDB, DSSP and Stride definitions for secondary structure elements and by using the consensus among those, we found 60,727 helices in 4,376 chains identified in all-α structures and 129,440 helices in 7,898 chains identified in all-α and α + β structures. For β-sheets, we identified 837,345 strands in 184,925 β-sheets located within 50,803 chains of all-β structures and 1,541,961 strands in 355,431 β-sheets located within 86,939 chains in all-β and α + β structures (data extracted on February 1, 2019). In this paper we would first like to address a full characterization of the nanoenvironment found at beta sheet locations and then compare those characteristics with the ones we already published for alpha helical secondary structure elements. For such characterization, we use here, as in our previous work about alpha helical nanoenvironments, set of STING protein structure descriptors. As in the previous work, we assume that we will be able to prove that there is a set of protein structure parameters/attributes/descriptors, which could fully describe the nanoenvironment around beta sheets and that appropriate statistically analysis will point out to significant changes in values for those parameters when compared for loci considered inside and outside defined secondary structure element. Clearly, while the univariate analysis is straightforward and intuitively understood, it is severely limited in coverage: it could be successfully applied at best in up to 25% of studied cases. The indication of the main descriptors for the specific secondary structure element (SSE) by means of the multivariate MANOVA test is the strong statistical tool for complete discrimination among the SSEs, and it revealed itself as the one with the highest coverage. The complete description of the nanoenvironment, by analogy, might be understood in terms of describing a key lock system, where all lock mini cylinders need to combine their elevation (controlled by a matching key) to open the lock. The main idea is as follows: a set of descriptors (cylinders in the key-lock example) must precisely combine their values (elevation) to form and maintain a specific secondary structure element nanoenvironment (a required condition for a key being able to open a lock).
- 30 December 2020
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 15 (12) , e0244315
Open AccessA comparison between internal protein nanoenvironments of α-helices and β-sheets
Secondary structure elements are generally found in almost all protein structures revealed so far. In general, there are more β-sheets than α helices found inside the protein structures. For example, considering the PDB, DSSP and Stride definitions for secondary structure elements and by using the consensus among those, we found 60,727 helices in 4,376 chains identified in all-α structures and 129,440 helices in 7,898 chains identified in all-α and α + β structures. For β-sheets, we identified 837,345 strands in 184,925 β-sheets located within 50,803 chains of all-β structures and 1,541,961 strands in 355,431 β-sheets located within 86,939 chains in all-β and α + β structures (data extracted on February 1, 2019). In this paper we would first like to address a full characterization of the nanoenvironment found at beta sheet locations and then compare those characteristics with the ones we already published for alpha helical secondary structure elements. For such characterization, we use here, as in our previous work about alpha helical nanoenvironments, set of STING protein structure descriptors. As in the previous work, we assume that we will be able to prove that there is a set of protein structure parameters/attributes/descriptors, which could fully describe the nanoenvironment around beta sheets and that appropriate statistically analysis will point out to significant changes in values for those parameters when compared for loci considered inside and outside defined secondary structure element. Clearly, while the univariate analysis is straightforward and intuitively understood, it is severely limited in coverage: it could be successfully applied at best in up to 25% of studied cases. The indication of the main descriptors for the specific secondary structure element (SSE) by means of the multivariate MANOVA test is the strong statistical tool for complete discrimination among the SSEs, and it revealed itself as the one with the highest coverage. The complete description of the nanoenvironment, by analogy, might be understood in terms of describing a key lock system, where all lock mini cylinders need to combine their elevation (controlled by a matching key) to open the lock. The main idea is as follows: a set of descriptors (cylinders in the key-lock example) must precisely combine their values (elevation) to form and maintain a specific secondary structure element nanoenvironment (a required condition for a key being able to open a lock).