Bioinformatics, 2023, 39(8), btad476 https://doi.org/10.1093/bioinformatics/btad476 Advance access publication 7 August 2023 Applications Note Structural bioinformatics TALAIA: a 3D visual dictionary for protein structures Mercè Alemany-Chavarria1, Jaime Rodrıguez-Guerra1, Jean-Didier Maréchal 1, * Downloaded from https://academic.oup.com/bioinformatics/article/39/8/btad476/7238214 by Withers user on 05 September 2023 1 Insilichem, Department of chemistry, Universitat Autònoma de Barcelona, Bellaterra 08193, Barcelona, Spain *Corresponding author. Insilichem, Department of chemistry, Universitat Autònoma de Barcelona, Bellaterra 08193, Barcelona, Spain. E-mail: jeandidier.marechal@uab.cat (J.-D.M.) Associate Editor: Lenore Cowen Abstract Motivation: Graphical analysis of the molecular structure of proteins can be very complex. Full-atom representations retain most geometric information but are generally crowded, and key structural patterns can be challenging to identify. Non-full-atom representations could be more in- structive on physicochemical aspects but be insufficiently detailed regarding shapes (e.g. entity beans-like models in coarse grain approaches) or simple properties of amino acids (e.g. representation of superficial electrostatic properties). In this work, we present TALAIA a visual dictionary that aims to provide another layer of structural representations. TALAIA offers a visual grammar that combines simple representations of amino acids while retaining their general geometry and physicochemi- cal properties. It uses unique objects, with differentiated shapes and colors to represent amino acids. It makes easier to spot crucial molecular in- formation, including patches of amino acids or key interactions between side chains. Most conventions used in TALAIA are standard in chemistry and biochemistry, so experimentalists and modelers can rapidly grasp the meaning of any TALAIA depiction. Results: We propose TALAIA as a tool that renders protein structures and encodes structure and physicochemical aspects as a simple visual grammar. The approach is fast, highly informative, and intuitive, allowing the identification of possible interactions, hydrophobic patches, and other characteristic structural features at first glance. The first implementation of TALAIA can be found at https://github.com/insilichem/talaia. Graphical abstract TALAIA is a 3D visual dictionary that simplifies the representation of amino acids while keeping their overall shape and relevant physicochemical properties. Received: February 7, 2023. Revised: April 30, 2023. Editorial Decision: May 27, 2023 C The Author(s) 2023. Published by Oxford University Press. V This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 2 Alemany-Chavarria et al. 1 Introduction conventions that are standards for molecular scientists. Regarding Molecular sciences are intrinsically related to our understanding colors, hydrophobic residues are depicted in green, positively of the three-dimensional (3D) nature of molecules. Graphic rep- charged groups in dark blue, and negatively charged in red. resentations, like those provided by computer software, allow Regarding forms, they tend to reproduce the most critical aspects us to perceive, describe, analyze, demonstrate, and communi- of the geometry of the amino acids side chains. Details are provided cate molecular knowledge, including morphological, dynamic below. properties, and reactivity. These representations stand on a se- Aliphatic groups are presented in spheric (Glycine, Alanine), ries of conventions and simplifications that relate to forms (e.g. ellipsoid (Valine, Leucine, Isoleucine, Methionine), or hexago- spheres for atoms, cylinders for bonds, etc.), colors (yellow for nal (Phenylalanine, Tryptophan) shapes. The elongation of the sulfur, red for oxygen, etc.), or textures (e.g. dot lines for weak ellipsoid and hexagons are functions of the dimensions of the interactions or depth of a bond for simple, double, or triple). side chain and take place along the central axis of the amino Downloaded from https://academic.oup.com/bioinformatics/article/39/8/btad476/7238214 by Withers user on 05 September 2023 Full-atom models are frequently challenging to interpret large acids. Methionine has a yellow ring around the ellipsoid’s end systems like proteins, while non-full-atom representations can where the sulfur atom is located. Neutral polar side chains, hide crucial in-depth information. For example, the ribbon repre- Serine, Threonine, and Cysteine are represented by a pyramid sentation of the backbone may obscure interactions of the pep- to indicate the direction of the polarity. The vertex points to tide chain’s oxygen and/or nitrogen atoms. Physicochemical the polar group, and the center is on the Ca. The pyramid tip is maps, like electrostatic ones, could be highly informative but yellow for Cys (the thiol group) and red for Ser and Thr (the miss the atomic details of the systems. A frequent solution con- hydroxyl group). Triangular prisms represent Glutamine and sists of allying both full-atom models with physicochemical Asparagine. The prism’s base is colored dark blue and red and maps. However, those hybrid representations could rapidly lead located over the polar atoms of the amide group. Tyrosine is to cluttered modeling and blurred visualization. Graphic repre- represented by a hexagon that overlaps with the residue’s ring, sentations that combine morphological (meaning maintaining and the hydroxyl moiety is colored red. most of the shape of the amino acids) and physicochemical infor- Histidine is depicted by an aquamarine pentagon overlap- mation are primarily absent from structural biology to date. The ping the residue’s ring when the residue is found in the neutral unique exception may be the Symbol Nomenclature for Glycans state. The color of the entire figure can change according to (Varki et al. 2015), a visual code proposed for saccharides. To its protonation state, dark blue when fully protonated and red some extent, the pipes-and-planks models of DNA or RNA em- when fully deprotonated. body this concept by representing backbone and nitrogen bases Positively charged residues Lysine and Arginine are represented in an intermediate graphical manner (Couch et al. 2006). by a blue rectangular box. The side of the cuboid where the amino Similarly, pharmacophores maintain the key structural 2D or 3D group (Lys) and the guanidyl group (Arg) lie is colored in a darker features of small molecules by blurring their complete atomic na- shade of blue. Aspartate and Glutamate side chains are depicted ture but maintain their fundamental physicochemical features with the exact figure as their amide forms, Asparagine and considering, eventually, in front of their closest environment Glutamine, and where the carboxylic group is found is depicted in (Schaller et al. 2020). a darker shade of red. Proline is a particular case since its struc- Here, we propose TALAIA as a visual dictionary for pro- tural characteristics and physicochemical properties can’t be in- teins. TALAIA provides a visual representation of amino acid cluded in any previous groups. Proline is therefore provided with a side chains that merges structural and some physicochemical unique code as a small gray cube centered on the Ca. This shape is information and could serve in inspecting protein structures, chosen due to the residue’s structural role. Along the same line, an evaluating interactions, or understanding the driving forces of orange star placed on top of the Ca was attributed to non- dynamical processes (e.g. a folding process). TALAIA can be standard amino acids. Specific objects for small molecules (ligands, applied to static structures and ensembles like trajectories solvents, or ions) in TALAIA are not considered yet. obtained from molecular dynamics simulations. 3.2 Illustrative cases For the sake of presenting TALAIA to the community, we se- 2 Availability and implementation lected two cases for which a reduced number of static pictures The first version of TALAIA is available as a UCSF Chimera are already informative. We selected the potassium channel of (Pettersen et al. 2004) extension and can be downloaded from the Mesorhizobium japonicum with PDB code 6eo1 (Kowal et al. GitHub repository at https://github.com/insilichem/talaia (see repos- 2018) and two snapshots from the conformational transition itory to install). Expansion toward other platforms is underway. of the oxaliplatin once bound to insulin obtained by molecu- In brief, the program reads the identity of the residues and lar dynamics (Sciortino et al. 2019). the atomic coordinates from the input file of the protein struc- Regarding the potassium channel (Fig. 1A), TALAIA allows us ture. Then it displays the corresponding graphical object on to identify subregions immediately: (1) a patch of negatively top of Chimera’s representation. These objects can overlap charged residues at the extracellular entrance of the pore, (2) small easily with the representation of the system’s secondary struc- residues and neutral polar residues at the core (the selectivity filter), ture and individual atoms (Fig. 1). This can be helped by set- and (3) a hydrophobic region formed mainly by S6 and C-linker ting the transparency of the figures manually. at the end of the transmembrane region. This representation allows fast recognition of a ring of positively charged arginine resi- dues where the cyclic nucleotide-binding domain binds near the 3 Description and illustrative cases membrane interface. Below that region, negatively charged resi- 3.1 Dictionary overview dues are located and mark the pore’s cytosolic end. TALAIA classifies amino acids into five families: hydrophobic, neu- Regarding the insulin–oxaliplatin system, Fig. 1B presents tral polar residues, positively and negatively charged residues, and two snapshots of the conformational transition observed in a proline (Table 1). Most of TALAIA’s objects are based on molecular dynamics simulation for one of the low-energy TALAIA 3 Downloaded from https://academic.oup.com/bioinformatics/article/39/8/btad476/7238214 by Withers user on 05 September 2023 Figure 1. (A) cAMP-bound potassium channel MloK1 (PDB: 6eo1). Residues inside the channel are represented with TALAIA. (B) Snapshots from the molecular dynamics’ simulation of the insulin–oxaliplatin complex with TALAIA’s representation for residues in 8 Å range from the drug, showing initial conformation (B.1) and final conformation (B.2). Table 1. Visual dictionary classification. complexes obtained by protein-ligand dockings (Sciortino the platinum ion with His26. TALAIA depiction allows us to et al. 2019). This complex results from the cleavage of the rapidly highlight that during the transition, the second coordi- Cys28–Cys7 disulfide bridge so that Cys28 can coordinate nation sphere interaction of the metal with Thr8 and Ser9 is 4 Alemany-Chavarria et al. lost (Fig. 1B.1). This allows the NTer part of the helix to un- Data availability fold, and finally, the aromatic part of the drug can reach a hy- The data underlying this article (code, installation procedure drophobic core formed by Ile10, Leu13, Ala35, Leu36, and tutorials) are available in https://github.com/insilichem/ Leu38, and Val39 (Fig. 1B.2). talaia. Guidance steps for illustrative cases including those To conclude, we believe that TALAIA provides an interest- produced in this manuscript are also available. ing complementary tool for the molecular inspection of bio- molecular systems by affording a visual code that preserves global geometric blueprints of amino acids that are physico- References chemically sound. Couch GS, Hendrix DK, Ferrin TE et al. Nucleic acid visualization with UCSF chimera. Nucleic Acids Res 2006;34:e29. Kowal J, Biyani N, Chami M et al. High-resolution cryoelectron micros- Downloaded from https://academic.oup.com/bioinformatics/article/39/8/btad476/7238214 by Withers user on 05 September 2023 Acknowledgements copy structure of the cyclic nucleotide-modulated potassium channel The authors thank the Spanish MINECO (project PID-2020- MloK1 in a lipid bilayer. Structure 2018;26:20–7.e3. 116861GB-I00). Pettersen EF, Goddard TD, Huang CC et al. UCSF chimera—a visuali- zation system for exploratory research and analysis. J Comput Chem 2004;25:1605–12. Conflict of interest Schaller D, Sribar D, Noonan T et al. Next generation 3D pharmaco- phore modeling. WIREs Comput Mol Sci 2020;10:e1468. None declared. Sciortino G, Sánchez-Aparicio J-E, Rodrıguez-Guerra Pedregal J et al. Computational insight into the interaction of oxaliplatin with insu- lin. Metallomics 2019;11:765–73. Varki A, Cummings RD, Aebi M et al. Symbol nomenclature for Funding graphical representations of glycans. Glycobiology 2015;25: Spanish MINECO project PID-2020-116861GB-I00. 1323–4.