Discovering Drugs with the Help of Machine Learning

Shown is a colour illustration of proteins in dozens of shapes, sizes and colours, on a white background. All the proteins have a grainy texture, but that is the only thing they have in common. Along the left are long, thin structures that look like threads of blue and purple twisted together. Next to it are what look like scattered orange breadcrumbs. In between is a tangle of thick red strands. Below is a pink sphere with a pattern that looks like blur flowers. To the right is something that looks like a hollow grey tube with different coloured strands wrapped around it in clumps. To the right are blue and purple clumps shaped like triangles, squares, rings and snowflakes. In the centre is a thick stripe that looks as if it is checkered in teal and purple. Next to it are two thick, lumpy strands of teal and purple twisted together. To the right is a thick, lumpy red strand, tangled at one end. Along the right edge is what looks like a long brown vine with green lumps along its length. In between are small to medium clumps in different shapes. They range in colour from purple and red to purple and teal.

Shown is a colour illustration of proteins during the four steps of folding. The title, “Protein Folding” is in bold letters across the top centre. Below are four small illustrations with descriptions. The first illustration shows a thin black strand spaced with brightly coloured dots, like beads in a necklace. The beads are labelled “Amino Acid,” and the strands between are labelled “Peptide Bond.” The description reads, “Primary Structure: The linear sequence of amino acids forms a chain.” In the second illustration, the coloured beads are now spaced out onto two new surfaces. The first is a bright blue ribbon, curled into a spiral. This is labelled “Alpha Helix.” The second is a long, pale blue sheet with an arrow pointing up at the top. This looks like it has been folded, accordion-style. This is labelled “Beta-pleated sheet.” The description below reads, “Secondary Structure: Short segments of the chain form into 3D structures that include alpha helices and beta sheets. In the third illustration, one bright blue spiral and two pale blue sheets from the previous image are piled on top of each other inside what looks like a fluffy white cloud. Thin black strands join the ends of each sheet to each other, as well as the spiral. The description reads “Tertiary Structure: The whole chain forms its 3D shape when segments fold up next to each other.” In the fourth illustration, two piles of sheets and spirals are next to each other in a larger white cloud. The description reads “Quaternary Structure: Often more than one chain come together to form a final protein structure.”

Shown is a colour illustration of the steps involved in two different approaches to protein drug discovery. The title, “Protein Drug Discovery” is in bold letters across the top. Below, the illustration is divided into two sections. The section on the left is subtitled “The Traditional Approach.” Starting at the top, the first illustration is a haystack labelled “Molecules to test.” A green arrow points from here down to three test tubes filled with red liquid. This is labelled “Wet lab experiments.” Another green arrow points from here down to the final illustration, a single sewing needle. This is labelled “Final drug design.” The section on the right is subtitled “The Generative Approach: Tell a computational model what you want and let it propose designs to test.” Around this is an oval-shaped diagram, with illustrations joined by blue arrows. At the top of the oval is a clipboard and pen. This is labelled “Specifications (wish list).” Arrows lead from here to an illustration that looks like two spiderwebs connected by gears. This is labelled “Computational model.” The next step has an illustration of three sewing needles. This is labelled “Candidate drug designs.” Next are three test tubes of red liquid labelled “Wet lab experiments.” The final illustration is a single sewing needle. This is labelled “Final drug design.” Arrows lead from here back up to the first illustration.

Shown is a colour computer rendering of protein molecules at the membrane of a nerve cell. Stretched horizontally across the illustration is a thick layer of tightly packed gold strands with small purple lumps along the top and bottom surfaces. This represents the phospholipid layer of a cell. Above this, small clumps of gold lumps float against a pale blue background. These represent small proteins. Below, two tall piles of purple lumps sit on top of the horizontal surface. These represent larger proteins embedded in the phospholipid layer. Below, more purple clumps float on a dark blue background, close to the bottom of the horizontal surface. These represent free-floating proteins. In the centre of the image, a few gold lumps and a few purple lumps have joined together across the horizontal layer. These represent proteins binding together in the membrane.