Arkadiusz Leniak
Arkadiusz Leniak has a total of 36 co-authors. Here are the top co-authors by number of shared publications:
- Wojciech Pietruś
Department of Medicinal Chemistry, Maj Institute of Pharmacology, Polish Academy of Sciences, Smetna 12, 31-343 Kraków, Poland
4 shared publications - Krzysztof Dubiel
Celon Pharma S.A., ul. Marymoncka 15, 05-152 Kazun Nowy, Poland
3 shared publications - Lidia Gurba-Bryśkiewicz
Celon Pharma S.A., ul. Marymoncka 15, 05-152 Kazun Nowy, Poland
3 shared publications - Maciej Wieczorek
Research and Development Center, Celon Pharma S.A., Marymoncka 15, 05-052 Kazuń Nowy, Poland
3 shared publications - Marcin Zagozda
Celon Pharma S.A., ul. Marymoncka 15, 05-152 Kazun Nowy, Poland
3 shared publications - Mariola Stypik
Celon Pharma S.A., ul. Marymoncka 15, 05-152 Kazun Nowy, Poland
3 shared publications
Latest Publications
- Open Access
- 16 July 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 66 (14) , 8318-8324
NMR-AI: An Open Platform for NMR-Enhanced Molecular Representations and Physicochemical Property Prediction
Accurate prediction of physicochemical properties is increasingly limited by an information ceiling of structure-only molecular descriptors. Here, predicted 1H|13C NMR chemical shifts are transformed into fixed-length NMR vectors and concatenated with ECFP4 to form the hybrid spectral–structural representation SpectraPRINTS, enabling direct evaluation of representational complementarity across logP, logS, and logD (pH 2.6, 7.4, and 10.5), as well as the most acidic and most basic pKas. With a fixed learning protocol, SpectraPRINT reduces error for lipophilicity- and solubility-related end points (up to 39% lower RMSE vs ECFP4), while no systematic gain is observed for the most acidic and most basic macroscopic pKa end points. The workflow is released as NMR-AI, a freely accessible web platform integrating NMR spectra prediction, descriptor construction, and property prediction, enabling interactive use and independent validation. The NMR-AI platform is accessible at https://cheminformaticsportal.if-pan.krakow.pl/.
- 16 July 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 66 (14) , 8318-8324
Open AccessNMR-AI: An Open Platform for NMR-Enhanced Molecular Representations and Physicochemical Property Prediction
Accurate prediction of physicochemical properties is increasingly limited by an information ceiling of structure-only molecular descriptors. Here, predicted 1H|13C NMR chemical shifts are transformed into fixed-length NMR vectors and concatenated with ECFP4 to form the hybrid spectral–structural representation SpectraPRINTS, enabling direct evaluation of representational complementarity across logP, logS, and logD (pH 2.6, 7.4, and 10.5), as well as the most acidic and most basic pKas. With a fixed learning protocol, SpectraPRINT reduces error for lipophilicity- and solubility-related end points (up to 39% lower RMSE vs ECFP4), while no systematic gain is observed for the most acidic and most basic macroscopic pKa end points. The workflow is released as NMR-AI, a freely accessible web platform integrating NMR spectra prediction, descriptor construction, and property prediction, enabling interactive use and independent validation. The NMR-AI platform is accessible at https://cheminformaticsportal.if-pan.krakow.pl/.
- Open Access
- 17 September 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (19) , 10323-10337
From NMR to AI: Fusing 1H and 13C Representations for Enhanced QSPR Modeling
The ability to predict log D directly from spectral patterns marks a conceptual shift in cheminformatics. In this work, we demonstrate that 1H and 13C NMR spectra, computationally generated from molecular structures and transformed into machine learning-compatible vectors, can approach and rival classical structure-based descriptors such as ECFP4 fingerprints in modeling the log D parameter. Through comprehensive benchmarking of nearly 70 models across seven algorithmic classes and three pH conditions, we show that concatenation of 1H and 13C NMR spectra offers the best trade-off between accuracy and efficiency. In the best case, a fused spectral CNN model achieved a root-mean-square error (RMSE) of 0.57 and a Q2 of 0.76 using a 400-dimensional input vector─closely matching the ECFP4 benchmark (RMSE 0.56, Q2 0.78) despite being five times smaller. These findings challenge the assumption that descriptor richness must come at the cost of dimensional complexity. SHAP-based analysis revealed modality-specific patterns: 13C regions linked to aromatic and carbonyl carbons (110-170 ppm) increased predicted log D, while 1H signals associated with polar groups, including OH, NH, amides, and ethers (2-4.5 and ∼8 ppm), reduced it. This positions NMR-based vectors as both interpretable and scalable alternatives to conventional fingerprints. By releasing a standalone graphical prediction tool based on our models, we make this paradigm practically accessible for real-world applications. This study establishes in silico-generated NMR spectra as valid and powerful descriptors in predictive modeling, paving the way for spectrum-driven approaches to drug discovery and property prediction.
- 17 September 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (19) , 10323-10337
Open AccessFrom NMR to AI: Fusing 1H and 13C Representations for Enhanced QSPR Modeling
The ability to predict log D directly from spectral patterns marks a conceptual shift in cheminformatics. In this work, we demonstrate that 1H and 13C NMR spectra, computationally generated from molecular structures and transformed into machine learning-compatible vectors, can approach and rival classical structure-based descriptors such as ECFP4 fingerprints in modeling the log D parameter. Through comprehensive benchmarking of nearly 70 models across seven algorithmic classes and three pH conditions, we show that concatenation of 1H and 13C NMR spectra offers the best trade-off between accuracy and efficiency. In the best case, a fused spectral CNN model achieved a root-mean-square error (RMSE) of 0.57 and a Q2 of 0.76 using a 400-dimensional input vector─closely matching the ECFP4 benchmark (RMSE 0.56, Q2 0.78) despite being five times smaller. These findings challenge the assumption that descriptor richness must come at the cost of dimensional complexity. SHAP-based analysis revealed modality-specific patterns: 13C regions linked to aromatic and carbonyl carbons (110-170 ppm) increased predicted log D, while 1H signals associated with polar groups, including OH, NH, amides, and ethers (2-4.5 and ∼8 ppm), reduced it. This positions NMR-based vectors as both interpretable and scalable alternatives to conventional fingerprints. By releasing a standalone graphical prediction tool based on our models, we make this paradigm practically accessible for real-world applications. This study establishes in silico-generated NMR spectra as valid and powerful descriptors in predictive modeling, paving the way for spectrum-driven approaches to drug discovery and property prediction.
- Open Access
Tuning the Biological Activity of PI3Kδ Inhibitor by the Introduction of a Fluorine Atom Using the Computational Workflow
As a member of the class I PI3K family, phosphoinositide 3-kinase δ (PI3Kδ) is an important signaling biomolecule that controls immune cell differentiation, proliferation, migration, and survival. It also represents a potential and promising therapeutic approach for the management of numerous inflammatory and autoimmune diseases. We designed and assessed the biological activity of new fluorinated analogues of CPL302415, taking into account the therapeutic potential of our selective PI3K inhibitor and fluorine introduction as one of the most frequently used modifications of a lead compound to further improve its biological activity. In this paper, we compare and evaluate the accuracy of our previously described and validated in silico workflow with that of the standard (rigid) molecular docking approach. The findings demonstrated that a properly fitted catalytic (binding) pocket for our chemical cores at the induced-fit docking (IFD) and molecular dynamics (MD) stages, along with QM-derived atomic charges, can be used for activity prediction to better distinguish between active and inactive molecules. Moreover, the standard approach seems to be insufficient to score the halogenated derivatives due to the fixed atomic charges, which do not consider the response and indictive effects caused by fluorine. The proposed computational workflow provides a computational tool for the rational design of novel halogenated drugs.
Open AccessTuning the Biological Activity of PI3Kδ Inhibitor by the Introduction of a Fluorine Atom Using the Computational Workflow
As a member of the class I PI3K family, phosphoinositide 3-kinase δ (PI3Kδ) is an important signaling biomolecule that controls immune cell differentiation, proliferation, migration, and survival. It also represents a potential and promising therapeutic approach for the management of numerous inflammatory and autoimmune diseases. We designed and assessed the biological activity of new fluorinated analogues of CPL302415, taking into account the therapeutic potential of our selective PI3K inhibitor and fluorine introduction as one of the most frequently used modifications of a lead compound to further improve its biological activity. In this paper, we compare and evaluate the accuracy of our previously described and validated in silico workflow with that of the standard (rigid) molecular docking approach. The findings demonstrated that a properly fitted catalytic (binding) pocket for our chemical cores at the induced-fit docking (IFD) and molecular dynamics (MD) stages, along with QM-derived atomic charges, can be used for activity prediction to better distinguish between active and inactive molecules. Moreover, the standard approach seems to be insufficient to score the halogenated derivatives due to the fixed atomic charges, which do not consider the response and indictive effects caused by fluorine. The proposed computational workflow provides a computational tool for the rational design of novel halogenated drugs.
- Open Access
- 30 July 2022
- Published by MDPI AG in Pharmaceuticals
- Vol. 15 (8) , 949
Design, Synthesis, and Development of pyrazolo[1,5-a]pyrimidine Derivatives as a Novel Series of Selective PI3Kδ Inhibitors: Part I—Indole Derivatives
Phosphoinositide 3-kinase δ (PI3Kδ), a member of the class I PI3K family, is an essential signaling biomolecule that regulates the differentiation, proliferation, migration, and survival of immune cells. The overactivity of this protein causes cellular dysfunctions in many human disorders, for example, inflammatory and autoimmune diseases, including asthma or chronic obstructive pulmonary disease (COPD). In this work, we designed and synthesized a new library of small-molecule inhibitors based on indol-4-yl-pyrazolo[1,5-a]pyrimidine with IC50 values in the low nanomolar range and high selectivity against the PI3Kδ isoform. CPL302253 (54), the most potent compound of all the structures obtained, with IC50 = 2.8 nM, is a potential future candidate for clinical development as an inhaled drug to prevent asthma.
- 30 July 2022
- Published by MDPI AG in Pharmaceuticals
- Vol. 15 (8) , 949
Open AccessDesign, Synthesis, and Development of pyrazolo[1,5-a]pyrimidine Derivatives as a Novel Series of Selective PI3Kδ Inhibitors: Part I—Indole Derivatives
Phosphoinositide 3-kinase δ (PI3Kδ), a member of the class I PI3K family, is an essential signaling biomolecule that regulates the differentiation, proliferation, migration, and survival of immune cells. The overactivity of this protein causes cellular dysfunctions in many human disorders, for example, inflammatory and autoimmune diseases, including asthma or chronic obstructive pulmonary disease (COPD). In this work, we designed and synthesized a new library of small-molecule inhibitors based on indol-4-yl-pyrazolo[1,5-a]pyrimidine with IC50 values in the low nanomolar range and high selectivity against the PI3Kδ isoform. CPL302253 (54), the most potent compound of all the structures obtained, with IC50 = 2.8 nM, is a potential future candidate for clinical development as an inhaled drug to prevent asthma.
- Open Access
- 27 July 2022
- Published by MDPI AG in Pharmaceuticals
- Vol. 15 (8) , 927
Design, Synthesis, and Development of Pyrazolo[1,5-a]pyrimidine Derivatives as a Novel Series of Selective PI3Kδ Inhibitors: Part II—Benzimidazole Derivatives
Phosphoinositide 3-kinase (PI3K) is the family of lipid kinases participating in vital cellular processes such as cell proliferation, growth, migration, or cytokines production. Due to the high expression of these proteins in many human cells and their involvement in metabolism regulation, normal embryogenesis, or maintaining glucose homeostasis, the inhibition of PI3K (especially the first class which contains four subunits: α, β, γ, δ) is considered to be a promising therapeutic strategy for the treatment of inflammatory and autoimmune diseases such as systemic lupus erythematosus (SLE) or multiple sclerosis. In this work, we synthesized a library of benzimidazole derivatives of pyrazolo[1,5-a]pyrimidine representing a collection of new, potent, active, and selective inhibitors of PI3Kδ, displaying IC50 values ranging from 1.892 to 0.018 μM. Among all compounds obtained, CPL302415 (6) showed the highest activity (IC50 value of 18 nM for PI3Kδ), good selectivity (for PI3Kδ relative to other PI3K isoforms: PI3Kα/δ = 79; PI3Kβ/δ = 1415; PI3Kγ/δ = 939), and promising physicochemical properties. As a lead compound synthesized on a relatively large scale, this structure is considered a potential future candidate for clinical trials in SLE treatment.
- 27 July 2022
- Published by MDPI AG in Pharmaceuticals
- Vol. 15 (8) , 927
Open AccessDesign, Synthesis, and Development of Pyrazolo[1,5-a]pyrimidine Derivatives as a Novel Series of Selective PI3Kδ Inhibitors: Part II—Benzimidazole Derivatives
Phosphoinositide 3-kinase (PI3K) is the family of lipid kinases participating in vital cellular processes such as cell proliferation, growth, migration, or cytokines production. Due to the high expression of these proteins in many human cells and their involvement in metabolism regulation, normal embryogenesis, or maintaining glucose homeostasis, the inhibition of PI3K (especially the first class which contains four subunits: α, β, γ, δ) is considered to be a promising therapeutic strategy for the treatment of inflammatory and autoimmune diseases such as systemic lupus erythematosus (SLE) or multiple sclerosis. In this work, we synthesized a library of benzimidazole derivatives of pyrazolo[1,5-a]pyrimidine representing a collection of new, potent, active, and selective inhibitors of PI3Kδ, displaying IC50 values ranging from 1.892 to 0.018 μM. Among all compounds obtained, CPL302415 (6) showed the highest activity (IC50 value of 18 nM for PI3Kδ), good selectivity (for PI3Kδ relative to other PI3K isoforms: PI3Kα/δ = 79; PI3Kβ/δ = 1415; PI3Kγ/δ = 939), and promising physicochemical properties. As a lead compound synthesized on a relatively large scale, this structure is considered a potential future candidate for clinical trials in SLE treatment.
- 18 July 2018
- Published by Wiley in Chemistry – An Asian Journal
- Vol. 14 (10) , 1763-1770
Building Molecular Complexity from Quinizarin: Conjoined Coumarins and Coronene Analogs
The double Knoevenagel condensation of 1,4‐dibenzoyloxyanthraquinone with methyl esters of arylacetic acids affords a series of compounds based upon a previously unknown 1,8‐dioxa‐benzo[e]pyrene‐2,7‐dione heterocyclic core. The aryl groups incorporated in the 3‐ and 6‐positions can be oxidatively coupled to the π‐expanded backbone to produce a further new heterocyclic core: 1,10‐dioxa‐dibenzo[dj]coronene‐2,9‐dione. The intriguing optical properties of these π‐expanded coumarin derivatives are discussed and rationalized through quantum chemical calculations. The broad absorption bands of 1,8‐dioxa‐benzo[e]pyrene‐2,7‐dione‐based dyes are attributed to both HOMO−1→LUMO and HOMO→LUMO transitions, which have a similar energy. Weakly coupled electron‐donating aryl substituents result in a moderate bathochromic shift of both the absorption and emission by 30–60 nm in toluene. The emissive properties of these compounds are in part determined by the oscillator strength of the main transition, lifetimes of the excited state, and by the energy match of the excited state with a triplet state of a similar energy. The 1,10‐dioxa‐dibenzo[dj]coronene‐2,9‐dione displays a much smaller Stokes shift, yet a markedly increased fluorescence quantum yield of 90 % owing to the increased rigidity compared with the 1,8‐dioxa‐benzo[e]pyrene‐2,7‐dione core.
- 18 July 2018
- Published by Wiley in Chemistry – An Asian Journal
- Vol. 14 (10) , 1763-1770
Building Molecular Complexity from Quinizarin: Conjoined Coumarins and Coronene Analogs
The double Knoevenagel condensation of 1,4‐dibenzoyloxyanthraquinone with methyl esters of arylacetic acids affords a series of compounds based upon a previously unknown 1,8‐dioxa‐benzo[e]pyrene‐2,7‐dione heterocyclic core. The aryl groups incorporated in the 3‐ and 6‐positions can be oxidatively coupled to the π‐expanded backbone to produce a further new heterocyclic core: 1,10‐dioxa‐dibenzo[dj]coronene‐2,9‐dione. The intriguing optical properties of these π‐expanded coumarin derivatives are discussed and rationalized through quantum chemical calculations. The broad absorption bands of 1,8‐dioxa‐benzo[e]pyrene‐2,7‐dione‐based dyes are attributed to both HOMO−1→LUMO and HOMO→LUMO transitions, which have a similar energy. Weakly coupled electron‐donating aryl substituents result in a moderate bathochromic shift of both the absorption and emission by 30–60 nm in toluene. The emissive properties of these compounds are in part determined by the oscillator strength of the main transition, lifetimes of the excited state, and by the energy match of the excited state with a triplet state of a similar energy. The 1,10‐dioxa‐dibenzo[dj]coronene‐2,9‐dione displays a much smaller Stokes shift, yet a markedly increased fluorescence quantum yield of 90 % owing to the increased rigidity compared with the 1,8‐dioxa‐benzo[e]pyrene‐2,7‐dione core.