Drug discovery typically consists of multiple steps, including identifying a target protein key to a disease's etiology, validating that interacting with this target could prevent symptoms or cure the disease, discovering a small molecule or biologic therapeutic to interact with it, and optimizing the candidate molecule through a complex landscape of required properties. Drug discovery related tasks often involve prediction and generation while considering multiple entities that potentially interact, which poses a challenge for typical AI models. For this purpose we present MAMMAL - Molecular Aligned Multi-Modal Architecture and Language - a method that we applied to create a versatile multi-task multi-align foundation model that learns from large-scale biological datasets (2 billion samples) across diverse modalities, including proteins, small molecules, and genes. We introduce a prompt syntax that supports a wide range of classification, regression, and generation tasks. It allows combining different modalities and entity types as inputs and/or outputs. Our model handles combinations of tokens and scalars and enables the generation of small molecules and proteins, property prediction, and transcriptomic lab test predictions. We evaluated the model on 11 diverse downstream tasks spanning different steps within a typical drug discovery pipeline, where it reaches new SOTA in 9 tasks and is comparable to SOTA in 2 tasks. This performance is achieved while using a unified architecture serving all tasks, in contrast to the original SOTA performance achieved using tailored architectures. The model code and pretrained weights are publicly available at https://github.com/BiomedSciAI/biomed-multi-alignment and https://huggingface.co/ibm/biomed.omics.bl.sm.ma-ted-458m.

Drug discovery is a complex, multi-stage process. Lead identification and lead optimization remain costly with low success-rates and computational methods play an important role in accelerating these tasks [1-3]. The prediction of a broad range of chemical and biological properties of candidate molecules is an essential component of screening and assessing molecules and data-driven, machine learning approaches have long aided in this process [4-6]. Molecular representations form the basis of machine learning models [2, 7], facilitating algorithmic and scientific advances in the field. However, learning useful and generalized latent representation is a hard problem due to limited amounts of labeled data, wide ranges of downstream tasks, vast chemical space, and large heterogeneity in molecular structures. Learning latent representations using unsupervised techniques is vital for such models to scale. Large language models (LLMs) have revolutionized other fields [8] and similar sequence-based foundation models have shown promise to learn molecular representations and be trainable on many downstream property prediction tasks [9-11]. A key advantage is that the transformer based architecture can learn in a self-supervised fashion to create a "pre-trained" molecular representation. The most direct application of LLM like transformers is facilitated by a sequence, text-based representation (e.g.

We present a simple, modular graph-based convolutional neural network that takes structural information from protein-ligand complexes as input to generate models for activity and binding mode prediction. Complex structures are generated by a standard docking procedure and fed into a dual-graph architecture that includes separate sub-networks for the ligand bonded topology and the ligand-protein contact map. This network division allows contributions from ligand identity to be distinguished from effects of protein-ligand interactions on classification. We show, in agreement with recent literature, that dataset bias drives many of the promising results on virtual screening that have previously been reported. However, we also show that our neural network is capable of learning from protein structural information when, as in the case of binding mode prediction, an unbiased dataset is constructed. We develop a deep learning model for binding mode prediction that uses docking ranking as input in combination with docking structures. This strategy mirrors past consensus models and outperforms the baseline docking program in a variety of tests, including on cross-docking datasets that mimic real-world docking use cases. Furthermore, the magnitudes of network predictions serve as reliable measures of model confidence