Abstract RNAs are a vast reservoir of untapped drug targets. Structure-based virtual screening (VS) identifies candidate molecules by leveraging binding site information, traditionally using molecular docking simulations. However, struggles to scale with large compound libraries and RNA Machine learning offers solution but remains underdeveloped for due limited data practical evaluations. We introduce data-driven VS pipeline tailored RNA, utilizing coarse-grained 3D modeling, synthetic...

Ligand-based drug design has recently benefited from the development of deep generative models. These models enable extensive explorations chemical space and provide a platform for molecular optimization. However, vast majority current methods does not leverage structure binding target, which potentiates small molecules plays key role in interaction. We propose an optimization pipeline that leverages complementary structure-based ligand-based methods. Instead performing docking on fixed...

RNA-small molecule binding is a key regulatory mechanism which can stabilize 3D structures and activate molecular functions. The discovery of RNA-targeting compounds thus current topic interest for novel therapies. Our work first attempt at bringing the scalability generalization abilities machine learning methods to problem RNA drug discovery, as well step towards understanding interactions drive specificity. tool, RNAmigos, builds encodes network representation predict likely ligands...

Abstract Motivation Therapeutic antibodies have emerged as a prominent class of new drugs due to their high specificity and ability bind several protein targets. Once an initial antibody has been identified, its design characteristics are refined using structural information, when it is available. Cryo-EM currently the most effective method obtain 3D structures. It relies on well-established methods process raw data into map, which may, however, be noisy contain artifacts. To fully interpret...

Abstract Predicting functional binding sites in proteins is crucial for understanding protein–protein interactions (PPIs) and identifying drug targets. While various computational approaches exist, many fail to assess PPI ligandability, which often involves conformational changes. We introduce InDeepNet, a web-based platform integrating InDeep, deep-learning model site prediction, with InDeepHolo, evaluates site’s propensity adopt ligand-bound (holo) conformation. InDeepNet provides an...

Protein-protein interactions (PPIs) are key elements in numerous biological pathways and the subject of a growing number drug discovery projects including against infectious diseases. Designing drugs on PPI targets remains difficult task requires extensive efforts to qualify given interaction as an eligible target. To this end, besides evident need determine role PPIs disease-associated their experimental characterization therapeutics targets, prediction capacity be bound by other protein...

Abstract Ligand-based drug design has recently benefited from the boost of deep generative models. These models enable extensive explorations chemical space, and provide a platform for molecular optimization. However, current state art methods do not leverage structure target, which is known to play key role in interaction. We propose an optimization pipeline that leverages complementary structure-based ligand-based methods. Instead performing docking on fixed bank, we iteratively select...

We implemented the Self-Organizing Maps algorithm running efficiently on GPUs, and also provide several clustering methods of resulting maps. scripts a use case to cluster macro-molecular conformations generated by molecular dynamics simulations.The method is available GitHub distributed as pip package.

Abstract Motivation RNA 3D motifs are recurrent substructures, modeled as networks of base pair interactions, which crucial for understanding structure–function relationships. The task automatically identifying such is computationally hard, and remains a key challenge in the field structural biology network analysis. State-of-the-art methods solve special cases motif problem by constraining variability occurrences motif, narrowing substructure search space. Results Here, we relax these...

RNA 3D motifs are recurrent substructures, modelled as networks of base pair interactions, which crucial for understanding structure-function relationships. The task automatically identifying such is computationally hard, and remains a key challenge in the field structural biology network analysis. State art methods solve special cases motif problem by constraining variability occurrences motif, narrowing substructure search space. Here, we relax these constraints posing finding graph...

Abstract RNAs constitute a vast reservoir of mostly untapped drug targets. Structure-based virtual screening (VS) methods screen large compound libraries for identifying promising candidate molecules by conditioning on binding site information. The classical approach relies molecular docking simulations. However, this strategy does not scale well with the size small molecule databases and number potential RNA Machine learning emerged as technology to resolve bottleneck. Efficient data-driven...

The current richness of sequence data needs efficient methodologies to display and analyze the complexity information in a compact readable manner. Traditionally, phylogenetic trees similarity networks have been used sequences protein families. These methods aim shed light on key computational biology problems such as classification functional inference. Here, we present new methodology, AlignScape, based self-organizing maps. AlignScape is applied three large families proteins: kinases...

Understanding the connection between complex structural features of RNA and biological function is a fundamental challenge in evolutionary studies design. However, building datasets 3D structures making appropriate modeling choices remains time-consuming lacks standardization. In this chapter, we describe use rnaglib, to train supervised unsupervised machine learning-based prediction models on structures.

Abstract As DNA sequencing technologies keep improving in scale and cost, there is a growing need to develop machine learning models analyze sequences, e.g., decipher regulatory signals from fragments bound by particular protein of interest. double helix made two complementary strands, fragment can be sequenced as equivalent, so-called Reverse Complement (RC) sequences nucleotides. To take into account this inherent symmetry the data facilitate learning. In sense, several authors have...

A bstract Motivation Protein-protein interactions (PPIs) are key elements in numerous biological pathways and the subject of a growing number drug discovery projects including against infectious diseases. Designing drugs on PPI targets remains difficult task requires extensive efforts to qualify given interaction as an eligible target. To this end, besides evident need determine role PPIs disease-associated their experimental characterization therapeutics targets, prediction capacity be...

Abstract Therapeutic antibodies have emerged as a prominent class of new drugs due to their high specificity and ability bind several protein targets. Once an initial antibody has been identified, optimization this hit compound follows based on the 3D structure, when available. Cryo-EM is currently most efficient method obtain such structures, supported by well-established methods that can transform raw data into potentially noisy map. These maps need be further interpreted inferring number,...

An essential aspect of learning from protein structures is the choice their representation as a geometric object (be it grid, graph, or surface), which conditions associated method. The performance given approach will then depend on both and its corresponding model. In this paper, we investigate representing proteins $\textit{surfaces embedded in 3D}$ evaluate within an established benchmark: atom3d. Our first finding that despite promising results, state-of-the-art surface-based approaches...

A bstract Motivation The binding of small molecules to RNAs is an important mechanism which can stabilize 3D structures or activate key molecular functions. To date, computational and experimental efforts toward molecule prediction have primarily focused on protein targets. Considering that a very large portion the genome transcribed into non-coding but only few regions are translated proteins, successful annotations RNA elements targeted by small-molecule would likely uncover vast...