PMCID
string
Title
string
Sentences
string
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Three replicate 10 × 10 combination matrices are generated for each drug pair, along with three replicates of each single-agent dose-response.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Twelve replicates each of cell death (orange) and vehicle (yellow) controls are included for normalization.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
d Overview of the Combocat analytical pipeline.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Raw data are mapped to corresponding combination matrices and subsequently normalized to percentage cell death using controls.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Synergy is then quantified using the Bliss independence model.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Results are summarized across the screen to identify top synergistic interactions.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
a Practical representation of drug synergy, where the combined effect of two drugs exceeds the expected effect (“E”) based on models like additivity or Bliss independence.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
b Examples of traditional high-throughput screens employing small (e.g., 3 × 3), sparse, or asymmetric (e.g., 2 × 7) matrix formats, which limit dose density.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
c Combocat’s streamlined approach for dense and reproducible drug combination screening.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The 384-well plate format contains Drugs 1 and 2 (represented by upper and lower triangles, respectively), with concentrations shown ranging from low (blue) to high (red).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Three replicate 10 × 10 combination matrices are generated for each drug pair, along with three replicates of each single-agent dose-response.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Twelve replicates each of cell death (orange) and vehicle (yellow) controls are included for normalization.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
d Overview of the Combocat analytical pipeline.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Raw data are mapped to corresponding combination matrices and subsequently normalized to percentage cell death using controls.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Synergy is then quantified using the Bliss independence model.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Results are summarized across the screen to identify top synergistic interactions.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Combination therapy is a promising frontier—but also a vast one whose comprehensive exploration has been limited by the sheer number of possible drug combinations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
With each additional agent tested, the number of total possible 2-drug combinations increases quadratically, making exhaustive experimental testing impractical.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
While large-scale single-agent screens have evaluated tens of thousands of compounds, drug combination screens have remained constrained by technical and resource limitations, preventing them from achieving comparable scale.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
These constraints often force researchers to choose between the number of combinations tested and the number of concentrations measured per drug, or dose density.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
This tradeoff can prove problematic, as dense measurements across a broad range of concentrations are crucial for capturing the nuanced dose-response landscapes needed to reliably detect synergy patterns.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Consequently, some of the largest reported combination studies have adopted smaller, sparser, or asymmetric matrix formats (Fig. 1b).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Advancements in robotic liquid handling technologies, such as acoustic dispensing, have revolutionized compound transfer, enabling more flexible and precise experimental designs.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Acoustic liquid handlers can dispense nanoliter volumes from any well of a source plate to any well of a destination plate without physical contact, enhancing throughput and minimizing resource consumption.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Despite the promise of these technological advances, their impact on drug combination screening has so far been limited by a lack of integrated experimental and analytical platforms that fully leverage their capabilities.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Here, we introduce Combocat, an open-source, end-to-end platform that integrates experimental and analytical workflows, underpinned by detailed protocols and documentation to enable reproducible, interpretable, and user-friendly drug combination screens.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Combocat operates in two complementary modes, each facilitating fast and efficient combination screening.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The “dense mode” measures all pairwise dose combinations in a dense 10 × 10 matrix format, tested in triplicate to ensure high fidelity and data quality.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Using dense mode, we generated a comprehensive reference dataset of over 290,000 unique combination measurements (spanning 806 unique drug combinations) across diverse drugs and cell types.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
While this approach is ideal for high-quality data-rich screens, scaling dense mode to thousands of combinations is resource-intensive.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
To address this, we developed a machine learning-assisted “sparse mode”.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Sparse mode reduces the number of direct experimental measurements by testing only the diagonal (10 dose pairs) of the 10 × 10 matrix and single-agent responses, then uses predictive modeling to obtain the remaining 90 values.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Trained on the dense mode reference dataset, the predictive models in sparse mode enable ultrahigh-throughput screening with minimal resources.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Notably, several other machine learning frameworks have been developed to predict synergy in different ways: TranSynergy leverages drug-target and gene-dependency features, SynToxProfiler computes integrated efficacy-toxicity-synergy surfaces, and comboFM uses tensor factorization to impute missing dose-response values.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
In contrast, Combocat’s sparse mode measures only the 10 diagonal dose pairs and recovers the complete 10 × 10 response landscape via an ensemble of 90 per-pair regression models.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
As proof of concept, we used sparse mode to screen 9045 drug combinations in the neuroblastoma cell line CHP-134, representing the largest number of unique combinations tested in a single cell line to date.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
By integrating advanced liquid handling technologies with machine learning into an open-source and user-friendly platform, Combocat offers a scalable, efficient solution for accelerating the discovery of drug combinations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
To address longstanding challenges in drug combination screens—such as limited dose density, poor interpretability, and operational complexity–we developed Combocat to integrate experimental and analytical workflows into a user-friendly framework.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The “dense mode” workflow facilitates comprehensive, dense measurements by systematically evaluating all pairwise dose combinations in a 10 × 10 matrix format (Fig. 1c).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
A customized acoustic liquid handling protocol prepares each 384-well plate with two drugs at ten concentrations each, including three replicate 10 × 10 combination matrices, single-agent dose-response curves, and internal controls (for example, twelve replicates each of cell death and vehicle controls).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
This design ensures robust normalization and reproducibility while reducing the time and effort typically associated with high-throughput combination screens.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Building on this experimental foundation, the analytical pipeline (Fig. 1d) maps raw data to their corresponding matrix positions, normalizes values to percentage cell death, and quantifies synergy using the Bliss independence model.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Results are then filtered and ranked, enabling researchers to rapidly identify top-performing combinations and visualize complete synergy landscapes without extensive post-processing.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
By uniting dense measurements with a streamlined analytical workflow, Combocat’s dense mode provides a high-fidelity framework for characterizing drug interactions in a reproducible and operationally straightforward manner.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Applying dense mode, we generated a reference dataset of 806 drug combinations comprising over 290,000 individual measurements in the 10 × 10 matrix format (Supplementary Data 1).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
This dataset was designed to evaluate a diverse range of drugs and cell types, ensuring generalizability across multiple experimental contexts.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Assay quality was high, as indicated by Z′ values (mean = 0.747; Supplementary Fig. 1) and low standard deviations in both single-agent and combination measurements across the screen (Fig. 2a, b).Fig.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
2Dense combination screening.a Distribution of single-agent standard deviations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The standard deviation is measured across each dose of all single-agent drugs across their respective three replicates.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
b Distribution of combination standard deviations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
For each drug combination, the standard deviation is measured across all 100 dose pairs and their respective three replicates.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
c Mean adjusted Bliss synergy (Blissadj.)
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
for all tested drug combinations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Each point represents the mean Blissadj.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
value across the 100 dose pairs per combination.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
SMX_TMP is highlighted as the top-ranking combination.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
d Dose-response curves for SMX (top) and TMP (bottom).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Four-parameter log-logistic curves were fit to the mean cell death values from the 10 doses.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Individual replicate values are represented by smaller gray dots.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Horizontal dashed lines correspond to 50% cell death.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Vertical dashed lines correspond to IC50 concentrations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
e Cell death combination matrix, represented as the mean of the three replicates.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
f Bliss synergy matrix calculated from (e).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
g Cell death for each single agent and their combination at doses corresponding to the maximum synergy.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The red dashed line indicates the expected response value under the Bliss model.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
h–k Like (d–g), for the combination of AZD1390 and CX-5461.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
a Distribution of single-agent standard deviations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The standard deviation is measured across each dose of all single-agent drugs across their respective three replicates.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
b Distribution of combination standard deviations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
For each drug combination, the standard deviation is measured across all 100 dose pairs and their respective three replicates.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
c Mean adjusted Bliss synergy (Blissadj.)
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
for all tested drug combinations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Each point represents the mean Blissadj.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
value across the 100 dose pairs per combination.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
SMX_TMP is highlighted as the top-ranking combination.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
d Dose-response curves for SMX (top) and TMP (bottom).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Four-parameter log-logistic curves were fit to the mean cell death values from the 10 doses.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Individual replicate values are represented by smaller gray dots.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Horizontal dashed lines correspond to 50% cell death.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Vertical dashed lines correspond to IC50 concentrations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
e Cell death combination matrix, represented as the mean of the three replicates.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
f Bliss synergy matrix calculated from (e).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
g Cell death for each single agent and their combination at doses corresponding to the maximum synergy.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The red dashed line indicates the expected response value under the Bliss model.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
h–k Like (d–g), for the combination of AZD1390 and CX-5461.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
During analysis, we identified occasional spurious single-agent measurements, a common challenge in high-throughput screens.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
These anomalies, often resulting from incomplete drug transfer or other technical issues, can distort synergy calculations by introducing outliers into the dataset (Supplementary Fig. 2a–d).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
To address this, we implemented a rigorous quality control (QC) pipeline that flags and adjusts anomalous data points using predefined thresholds for variability, dose-response residuals, and monotonicity (Supplementary Fig. 2e–g; see “Methods” section for details).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Flagged values can be excluded from subsequent synergy calculations, yielding an “adjusted Bliss synergy” (Blissadj.)
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
that omits these measurements.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
This QC approach ensures accurate synergy quantification by minimizing the impact of outliers.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
To evaluate the platform’s capacity to identify and replicate known synergistic interactions, we first examined the top-ranked combination in our dataset: sulfamethoxazole (SMX) and trimethoprim (TMP) in E. coli cells (Fig. 2c).
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
This pair is a well-established example of synergistic drug behavior, widely used to treat various bacterial infections.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Both SMX and TMP displayed sigmoidal dose-dependent relationships (Fig. 2d), and when combined, exhibited strong synergy (Fig. 2e–g), aligning with established expectations.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
We further evaluated an additional known example of synergy included in our screen—the combination of AZD1390 and CX-5461—which has been reported as synergistic in Neuroblastoma cells.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The single-agent responses and strong synergy patterns (Fig. 2h–k) closely aligned to reported findings.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
Collectively, these examples highlight Combocat’s ability to faithfully reproduce known synergistic interactions and provide a framework for understanding complex combination landscapes.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
While dense mode offers high-quality, data-rich combination measurements, scaling it to thousands of combinations can become resource-intensive.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
To address this, we developed a “sparse mode” workflow, which minimizes direct experimental measurements while still capturing synergy information between drug pairs.
PMC12705714
An open-source screening platform accelerates discovery of drug combinations
The sparse mode assay is miniaturized into a 1536-well format and separates the single agents and combinations into their own plates (Supplementary Fig. 3a, b).