Douglas R. Houston
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Douglas R. Houston has a total of 295 co-authors. Here are the top co-authors by number of shared publications:
- Vincent Blay
Department of Microbiology and Environmental Toxicology, University of California Santa Cruz, Santa Cruz, CA 95064, USA
10 shared publications - Boyang Ni
University of Edinburgh
4 shared publications - Ted R. Hupp
University of Edinburgh
4 shared publications - Anders Hogner
Medicinal Chemistry, Research and Early Development, Cardiovascular, Renal and Metabolism (CVRM), BioPharmaceuticals R&D
3 shared publications - Dakota Treleaven
Conscience Medicines Network
3 shared publications - Elisa Gibson
Structural Genomics Consortium
3 shared publications
Latest Publications
- 13 November 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
Critical Assessments of Computational Hit-finding Experiments Challenge #4: Targeting the Tyrosine Kinase Binding Domain of the E3 Ligase Casitas B-Lineage Lymphoma Proto-Oncogene B, an Immuno-Oncology Target
The critical assessments of computational hit-finding experiments (CACHE) are prospective benchmarking exercises where small molecule ligands predicted for a nominated protein of interest are procured and tested experimentally. In CACHE #4, 23 participating teams each selected up to 100 compounds designed to bind the E3 ligase CBLB, an immuno-oncology target. The organizers released the first public structure of CBLB in complex with a known ligand at the outset of the challenge to enable the discovery of chemically novel molecules. Out of the 1688 compounds collectively predicted, ten compounds from nine participants showed some sign of activity (primary hits), but only two were convincingly confirmed (validated hits) across multiple biophysical binding assays and in subsequent analogs, and only one of them was chemically novel. The apparent difficulty of finding novel CBLB ligands, previously reflected by the chemical similarity of inhibitors from the patent literature, suggests that potent CBLB hit molecules were either rare in the screened commercial libraries, or that computational methods may have struggled to retrieve presumably present but weak ligands, or that the conformational dynamics of the binding site was not sufficiently accounted for. The successful method was strikingly atypical. Selection from a chemical binding similarity classifier followed by a random forest binding affinity predictor trained on inhibitors from the patent literature was refined using a diffusion-model binding pose predictor. The road to an ML-driven breakthrough in computational hit-finding can take unexpected turns. Which of these turns reveal novel horizons and which are mere distractions will best be investigated and answered collectively. The discovery in CACHE #4 of a novel series of CBLB ligands is not a trivial achievement and may inform medicinal chemistry efforts toward clinical candidates that complement immune checkpoint inhibition to elicit antitumor immune response in cancer patients.
- 13 November 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
Critical Assessments of Computational Hit-finding Experiments Challenge #4: Targeting the Tyrosine Kinase Binding Domain of the E3 Ligase Casitas B-Lineage Lymphoma Proto-Oncogene B, an Immuno-Oncology Target
The critical assessments of computational hit-finding experiments (CACHE) are prospective benchmarking exercises where small molecule ligands predicted for a nominated protein of interest are procured and tested experimentally. In CACHE #4, 23 participating teams each selected up to 100 compounds designed to bind the E3 ligase CBLB, an immuno-oncology target. The organizers released the first public structure of CBLB in complex with a known ligand at the outset of the challenge to enable the discovery of chemically novel molecules. Out of the 1688 compounds collectively predicted, ten compounds from nine participants showed some sign of activity (primary hits), but only two were convincingly confirmed (validated hits) across multiple biophysical binding assays and in subsequent analogs, and only one of them was chemically novel. The apparent difficulty of finding novel CBLB ligands, previously reflected by the chemical similarity of inhibitors from the patent literature, suggests that potent CBLB hit molecules were either rare in the screened commercial libraries, or that computational methods may have struggled to retrieve presumably present but weak ligands, or that the conformational dynamics of the binding site was not sufficiently accounted for. The successful method was strikingly atypical. Selection from a chemical binding similarity classifier followed by a random forest binding affinity predictor trained on inhibitors from the patent literature was refined using a diffusion-model binding pose predictor. The road to an ML-driven breakthrough in computational hit-finding can take unexpected turns. Which of these turns reveal novel horizons and which are mere distractions will best be investigated and answered collectively. The discovery in CACHE #4 of a novel series of CBLB ligands is not a trivial achievement and may inform medicinal chemistry efforts toward clinical candidates that complement immune checkpoint inhibition to elicit antitumor immune response in cancer patients.
- Open Access
Evaluation of Trypanosoma brucei Phosphofructokinase Allosteric Inhibition: An In-Silico Study
Human African trypanosomiasis, caused by a protozoan parasite Trypanosoma brucei , is a neglected tropical disease for which well-tolerated, conveniently administered, and highly efficacious medicines are still missing. Previously, T. brucei Phosphofructokinase was targeted by small-molecule inhibitor development efforts. This approach has shown promise both in vitro and in vivo . In this study, we have used these wet-lab results, evaluated the compounds already characterised by Molecular Dynamics simulations, found relationships between in silico and wet-lab data and used these observations to evaluate compounds that we selected through several different approaches of virtual screens. We observed that inhibitor-ATP interactions are highly predictive of the inhibitory activity. Several compounds selected through virtual screens have outperformed previously characterised compounds.
Open AccessEvaluation of Trypanosoma brucei Phosphofructokinase Allosteric Inhibition: An In-Silico Study
Human African trypanosomiasis, caused by a protozoan parasite Trypanosoma brucei , is a neglected tropical disease for which well-tolerated, conveniently administered, and highly efficacious medicines are still missing. Previously, T. brucei Phosphofructokinase was targeted by small-molecule inhibitor development efforts. This approach has shown promise both in vitro and in vivo . In this study, we have used these wet-lab results, evaluated the compounds already characterised by Molecular Dynamics simulations, found relationships between in silico and wet-lab data and used these observations to evaluate compounds that we selected through several different approaches of virtual screens. We observed that inhibitor-ATP interactions are highly predictive of the inhibitory activity. Several compounds selected through virtual screens have outperformed previously characterised compounds.
- 24 February 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 66 (5) , 2735-2752
UniDock-Pro: A Unified GPU-Accelerated Platform for High-Throughput Structure-Based, Ligand-Based, and Synergistic Hybrid Virtual Screening
The screening of ultralarge chemical libraries requires high-throughput computational tools that can efficiently exploit all available structural and chemical information. We present UniDock-Pro, a unified platform based on the GPU-accelerated Uni-Dock architecture that integrates structure-based virtual screening (SBVS), ligand-based virtual screening (LBVS), and a novel hybrid virtual screening mode. By capitalizing on inter-ligand (batch) parallelism, UniDock-Pro achieves substantial throughput gains, enabling the processing of millions of compounds per day on a single GPU. We significantly enhanced the LBVS methodology by implementing a smooth, Lennard-Jones-like potential optimized for gradient-based search, which replaces the rugged recursive model employed in our previous work. This optimization yields a substantial 2.42-fold improvement in early enrichment (EF1%) over the legacy AutoDock-SS on the DUDE-Z benchmark. The Hybrid mode combines complementary information by integrating receptor- and ligand-derived grid maps on-the-fly during the conformational search, delivering strong early enrichment on DUDE-Z and competitive performance on the more stringent VSDS-vd TrueDecoy benchmark. To understand the mechanisms underlying this synergy, we introduce Force Field Complementarity Analysis (FFCA), a method for quantifying the spatial alignment between receptor and ligand force fields. UniDock-Pro offers a robust, versatile, and highly efficient solution for accelerating drug discovery campaigns across the expansive modern chemical space.
- 24 February 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 66 (5) , 2735-2752
UniDock-Pro: A Unified GPU-Accelerated Platform for High-Throughput Structure-Based, Ligand-Based, and Synergistic Hybrid Virtual Screening
The screening of ultralarge chemical libraries requires high-throughput computational tools that can efficiently exploit all available structural and chemical information. We present UniDock-Pro, a unified platform based on the GPU-accelerated Uni-Dock architecture that integrates structure-based virtual screening (SBVS), ligand-based virtual screening (LBVS), and a novel hybrid virtual screening mode. By capitalizing on inter-ligand (batch) parallelism, UniDock-Pro achieves substantial throughput gains, enabling the processing of millions of compounds per day on a single GPU. We significantly enhanced the LBVS methodology by implementing a smooth, Lennard-Jones-like potential optimized for gradient-based search, which replaces the rugged recursive model employed in our previous work. This optimization yields a substantial 2.42-fold improvement in early enrichment (EF1%) over the legacy AutoDock-SS on the DUDE-Z benchmark. The Hybrid mode combines complementary information by integrating receptor- and ligand-derived grid maps on-the-fly during the conformational search, delivering strong early enrichment on DUDE-Z and competitive performance on the more stringent VSDS-vd TrueDecoy benchmark. To understand the mechanisms underlying this synergy, we introduce Force Field Complementarity Analysis (FFCA), a method for quantifying the spatial alignment between receptor and ligand force fields. UniDock-Pro offers a robust, versatile, and highly efficient solution for accelerating drug discovery campaigns across the expansive modern chemical space.
- Open Access
- 8 November 2025
- Published by Elsevier in ImmunoInformatics
- Vol. 20
DoggifAI: A transformer based approach for antibody caninisation
- 8 November 2025
- Published by Elsevier in ImmunoInformatics
- Vol. 20
Open AccessDoggifAI: A transformer based approach for antibody caninisation
- Open Access
- 20 August 2025
- Published by Wiley in Advanced Science
- Vol. 12 (42) , e08318
SCORCH2: A Generalized Heterogeneous Consensus Model for High‐Enrichment Interaction‐Based Virtual Screening
The discovery of effective therapeutics remains a complex, costly, and time‐consuming endeavor, characterized by high failure rates and significant resource investments. A central bottleneck in early‐stage drug discovery is identifying suitable hit compounds with moderate affinity for known biological targets. Although advancements occur, current in silico virtual screening methods are subject to limitations, including model overfitting, data bias, and constrained interpretability in their predictive processes. In this study, we present SCORCH2, a machine learning‐based framework designed to simultaneously enhance the performance and interpretability of virtual screening by leveraging interaction features. Comparing with its predecessor SCORCH, SCORCH2 exhibits superior predictive accuracy and generalizability across a wide range of biological targets. Importantly, SCORCH2 demonstrates robust hit identification capabilities on previously unseen targets, indicating strong transferability. Furthermore, SCORCH2 obviates the need for meticulous docking pose selection, streamlining the screening process. These advances highlight the potential of SCORCH2 as a valuable tool in accelerating drug discovery campaigns.
- 20 August 2025
- Published by Wiley in Advanced Science
- Vol. 12 (42) , e08318
Open AccessSCORCH2: A Generalized Heterogeneous Consensus Model for High‐Enrichment Interaction‐Based Virtual Screening
The discovery of effective therapeutics remains a complex, costly, and time‐consuming endeavor, characterized by high failure rates and significant resource investments. A central bottleneck in early‐stage drug discovery is identifying suitable hit compounds with moderate affinity for known biological targets. Although advancements occur, current in silico virtual screening methods are subject to limitations, including model overfitting, data bias, and constrained interpretability in their predictive processes. In this study, we present SCORCH2, a machine learning‐based framework designed to simultaneously enhance the performance and interpretability of virtual screening by leveraging interaction features. Comparing with its predecessor SCORCH, SCORCH2 exhibits superior predictive accuracy and generalizability across a wide range of biological targets. Importantly, SCORCH2 demonstrates robust hit identification capabilities on previously unseen targets, indicating strong transferability. Furthermore, SCORCH2 obviates the need for meticulous docking pose selection, streamlining the screening process. These advances highlight the potential of SCORCH2 as a valuable tool in accelerating drug discovery campaigns.
- Open Access
- 19 August 2025
- Published by American Chemical Society (ACS) in ChemRxiv
UniDock-Pro: A Unified GPU-Accelerated Platform for High-Throughput Structure-Based, Ligand-Based, and Synergistic Hybrid Virtual Screening
The imperative to screen ultra-large chemical libraries necessitates high-throughput computational tools capable of efficiently leveraging all available structural and chemical information. We introduce UniDock-Pro, a unified platform built upon the GPU-accelerated Uni-Dock architecture, which integrates structure-based (SBVS), ligand-based (LBVS), and a novel Hybrid virtual screening mode. By capitalizing on inter-ligand (batch) parallelism, UniDock-Pro achieves massive throughput gains, processing millions of compounds per day on a single GPU. We significantly enhanced the LBVS methodology by implementing a smooth, Lennard-Jones-like potential optimized for gradient-based search, replacing the rugged recursive model used in prior tools. This optimization yields a substantial 2.45-fold improvement in early enrichment ($\mathrm{EF}_{1\%}$) over the legacy AutoDock-SS on the DUDE-Z benchmark. The Hybrid mode synergizes orthogonal information by integrating receptor- and ligand-derived grid maps on-the-fly during the conformational search, consistently achieving the highest overall enrichment across diverse benchmarks. To elucidate the mechanisms underlying this synergy, we introduce Force Field Complementarity Analysis (FFCA), a method for quantifying the spatial alignment between receptor and ligand force fields. UniDock-Pro provides a robust, versatile, and highly efficient solution for accelerating drug discovery campaigns across the vast modern chemical space.
- 19 August 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessUniDock-Pro: A Unified GPU-Accelerated Platform for High-Throughput Structure-Based, Ligand-Based, and Synergistic Hybrid Virtual Screening
The imperative to screen ultra-large chemical libraries necessitates high-throughput computational tools capable of efficiently leveraging all available structural and chemical information. We introduce UniDock-Pro, a unified platform built upon the GPU-accelerated Uni-Dock architecture, which integrates structure-based (SBVS), ligand-based (LBVS), and a novel Hybrid virtual screening mode. By capitalizing on inter-ligand (batch) parallelism, UniDock-Pro achieves massive throughput gains, processing millions of compounds per day on a single GPU. We significantly enhanced the LBVS methodology by implementing a smooth, Lennard-Jones-like potential optimized for gradient-based search, replacing the rugged recursive model used in prior tools. This optimization yields a substantial 2.45-fold improvement in early enrichment ($\mathrm{EF}_{1\%}$) over the legacy AutoDock-SS on the DUDE-Z benchmark. The Hybrid mode synergizes orthogonal information by integrating receptor- and ligand-derived grid maps on-the-fly during the conformational search, consistently achieving the highest overall enrichment across diverse benchmarks. To elucidate the mechanisms underlying this synergy, we introduce Force Field Complementarity Analysis (FFCA), a method for quantifying the spatial alignment between receptor and ligand force fields. UniDock-Pro provides a robust, versatile, and highly efficient solution for accelerating drug discovery campaigns across the vast modern chemical space.
- 20 June 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (13) , 6884-6898
CACHE Challenge #2: Targeting the RNA Site of the SARS-CoV-2 Helicase Nsp13
A critical assessment of computational hit-finding experiments (CACHE) challenge was conducted to predict ligands for the SARS-CoV-2 Nsp13 helicase RNA binding site, a highly conserved COVID-19 target. Twenty-three participating teams comprised of computational chemists and data scientists used protein structure and data from fragment-screening paired with advanced computational and machine learning methods to each predict up to 100 inhibitory ligands. Across all teams, 1957 compounds were predicted and were subsequently procured from commercial catalogs for biophysical assays. Of these compounds, 0.7% were confirmed to bind to Nsp13 in a surface plasmon resonance assay. The six best-performing computational workflows used fragment growing, active learning, or conventional virtual screening with and without complementary deep-learning scoring functions. Follow-up functional assays resulted in identification of two compound scaffolds that bound Nsp13 with a Kd below 10 μM and inhibited in vitro helicase activity. Overall, CACHE #2 participants were successful in identifying hit compound scaffolds targeting Nsp13, a central component of the coronavirus replication-transcription complex. Computational design strategies recurrently successful across the first two CACHE challenges include linking or growing docked or crystallized fragments and docking small and diverse libraries to train ultrafast machine-learning models. The CACHE #2 competition reveals how crowd-sourcing ligand prediction efforts using a distinct array of approaches followed with critical biophysical assays can result in novel lead compounds to advance drug discovery efforts.
- 20 June 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (13) , 6884-6898
CACHE Challenge #2: Targeting the RNA Site of the SARS-CoV-2 Helicase Nsp13
A critical assessment of computational hit-finding experiments (CACHE) challenge was conducted to predict ligands for the SARS-CoV-2 Nsp13 helicase RNA binding site, a highly conserved COVID-19 target. Twenty-three participating teams comprised of computational chemists and data scientists used protein structure and data from fragment-screening paired with advanced computational and machine learning methods to each predict up to 100 inhibitory ligands. Across all teams, 1957 compounds were predicted and were subsequently procured from commercial catalogs for biophysical assays. Of these compounds, 0.7% were confirmed to bind to Nsp13 in a surface plasmon resonance assay. The six best-performing computational workflows used fragment growing, active learning, or conventional virtual screening with and without complementary deep-learning scoring functions. Follow-up functional assays resulted in identification of two compound scaffolds that bound Nsp13 with a Kd below 10 μM and inhibited in vitro helicase activity. Overall, CACHE #2 participants were successful in identifying hit compound scaffolds targeting Nsp13, a central component of the coronavirus replication-transcription complex. Computational design strategies recurrently successful across the first two CACHE challenges include linking or growing docked or crystallized fragments and docking small and diverse libraries to train ultrafast machine-learning models. The CACHE #2 competition reveals how crowd-sourcing ligand prediction efforts using a distinct array of approaches followed with critical biophysical assays can result in novel lead compounds to advance drug discovery efforts.
- Open Access
DoggifAI: a transformer based approach for antibody caninisation
Antibody translation across species offers a compelling strategy to extend the vast and expensive investments in human therapeutic antibodies to veterinary oncology, with applications in both veterinary medicine and comparative oncology. While precise, low-immunogenic treatments are essential for canine cancer care, traditional species conversion methods rely on ad hoc bioinformatics modifications. These methods often implicitly decouple the framework (FR) and complementarity-determining regions (CDRs), ignoring how structural changes in FRs can affect the conformation and function of CDRs. This can compromise binding specificity and require costly high-throughput in vitro screening. To address this, we present DoggifAI, a transformer model that translates non-canine antibody sequences into canine ones by generating species-appropriate framework regions (FRs) based on desired CDRs. This allows the model to better preserve structural compatibility between FRs and CDRs. The model is pretrained in a T5-style text-to-text denoising task on a large multispecies antibody dataset, which allows further finetuning on a much smaller species-specific dataset. DoggifAI generates highly canine-like antibodies and shows promising results in preserving binding specificity. To support further progress in this field, we also release a curated dataset of over 430,000 unique canine antibody chain sequences, significantly expanding the public sequence repertoire.
Open AccessDoggifAI: a transformer based approach for antibody caninisation
Antibody translation across species offers a compelling strategy to extend the vast and expensive investments in human therapeutic antibodies to veterinary oncology, with applications in both veterinary medicine and comparative oncology. While precise, low-immunogenic treatments are essential for canine cancer care, traditional species conversion methods rely on ad hoc bioinformatics modifications. These methods often implicitly decouple the framework (FR) and complementarity-determining regions (CDRs), ignoring how structural changes in FRs can affect the conformation and function of CDRs. This can compromise binding specificity and require costly high-throughput in vitro screening. To address this, we present DoggifAI, a transformer model that translates non-canine antibody sequences into canine ones by generating species-appropriate framework regions (FRs) based on desired CDRs. This allows the model to better preserve structural compatibility between FRs and CDRs. The model is pretrained in a T5-style text-to-text denoising task on a large multispecies antibody dataset, which allows further finetuning on a much smaller species-specific dataset. DoggifAI generates highly canine-like antibodies and shows promising results in preserving binding specificity. To support further progress in this field, we also release a curated dataset of over 430,000 unique canine antibody chain sequences, significantly expanding the public sequence repertoire.
- Open Access
SCORCH2: a generalised heterogeneous consensus model for high-enrichment interaction-based virtual screening
The discovery of effective therapeutics remains a complex, costly, and time-consuming endeavor, characterized by high failure rates and significant resource investments. A central bottleneck in early-stage drug discovery is identifying suitable hit compounds with moderate affinity for known biological targets. Although advancements have been made, current in silico virtual screening methods are subject to limitations, including model overfitting, data bias, and constrained interpretability in their predictive processes. In this study, we present SCORCH2, a machine learning-based framework designed to enhance both the performance and interpretability of virtual screening by leveraging interaction features. Compared with its predecessor SCORCH, SCORCH2 exhibits superior predictive accuracy and generalizability across a wide range of biological targets. Importantly, SCORCH2 demonstrates robust hit identification capabilities on previously unseen targets, indicating strong transferability. These results highlight the potential of SCORCH2 as a valuable tool in accelerating drug discovery, offering reliable predictive capabilities while improving the interpretability of virtual screening models.
Open AccessSCORCH2: a generalised heterogeneous consensus model for high-enrichment interaction-based virtual screening
The discovery of effective therapeutics remains a complex, costly, and time-consuming endeavor, characterized by high failure rates and significant resource investments. A central bottleneck in early-stage drug discovery is identifying suitable hit compounds with moderate affinity for known biological targets. Although advancements have been made, current in silico virtual screening methods are subject to limitations, including model overfitting, data bias, and constrained interpretability in their predictive processes. In this study, we present SCORCH2, a machine learning-based framework designed to enhance both the performance and interpretability of virtual screening by leveraging interaction features. Compared with its predecessor SCORCH, SCORCH2 exhibits superior predictive accuracy and generalizability across a wide range of biological targets. Importantly, SCORCH2 demonstrates robust hit identification capabilities on previously unseen targets, indicating strong transferability. These results highlight the potential of SCORCH2 as a valuable tool in accelerating drug discovery, offering reliable predictive capabilities while improving the interpretability of virtual screening models.
- Open Access
- 5 March 2025
- Published by American Chemical Society (ACS) in ChemRxiv
CACHE Challenge #2: Targeting the RNA Site of the SARS-CoV-2 Helicase Nsp13
A critical assessment of computational hit finding experiments (CACHE) challenge was conducted to predict ligands for the SARS-CoV-2 Nsp13 helicase RNA binding site, a highly conserved COVID-19 target. Twenty-three participating teams comprised of computational chemists and data scientists used protein structure and data from fragment-screening paired with advanced computational and machine learning methods to each predict up to 100 inhibitory ligands. Across all teams, 1957 compounds were predicted and were subsequently procured from commercial catalogs for biophysical assays. Of these compounds, 0.7% were confirmed to bind to Nsp13 in a surface plasmon resonance assay. The six best performing computational workflows used fragment growing, active learning, or conventional virtual screening with and without complementary deep-learning scoring functions. Follow-up functional assays resulted in identification of two compound scaffolds that bound Nsp13 with a Kd below 10 µM and inhibited in vitro helicase activity. Overall, the CACHE #2 was successful in identifying hit compound scaffolds targeting Nsp13, a central component of the coronavirus replication-transcription complex. Computational design strategies recurrently successful across the first two CACHE challenges include linking or growing docked or crystallized fragments and docking small and diverse libraries to train ultra-fast machine-learning models. The CACHE#2 competition reveals how crowd-sourcing ligand prediction efforts using a distinct array of approaches followed with critical biophysical assays can result in novel lead compounds to advance drug discovery efforts.
- 5 March 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessCACHE Challenge #2: Targeting the RNA Site of the SARS-CoV-2 Helicase Nsp13
A critical assessment of computational hit finding experiments (CACHE) challenge was conducted to predict ligands for the SARS-CoV-2 Nsp13 helicase RNA binding site, a highly conserved COVID-19 target. Twenty-three participating teams comprised of computational chemists and data scientists used protein structure and data from fragment-screening paired with advanced computational and machine learning methods to each predict up to 100 inhibitory ligands. Across all teams, 1957 compounds were predicted and were subsequently procured from commercial catalogs for biophysical assays. Of these compounds, 0.7% were confirmed to bind to Nsp13 in a surface plasmon resonance assay. The six best performing computational workflows used fragment growing, active learning, or conventional virtual screening with and without complementary deep-learning scoring functions. Follow-up functional assays resulted in identification of two compound scaffolds that bound Nsp13 with a Kd below 10 µM and inhibited in vitro helicase activity. Overall, the CACHE #2 was successful in identifying hit compound scaffolds targeting Nsp13, a central component of the coronavirus replication-transcription complex. Computational design strategies recurrently successful across the first two CACHE challenges include linking or growing docked or crystallized fragments and docking small and diverse libraries to train ultra-fast machine-learning models. The CACHE#2 competition reveals how crowd-sourcing ligand prediction efforts using a distinct array of approaches followed with critical biophysical assays can result in novel lead compounds to advance drug discovery efforts.