1<!doctype html><html lang=en-us> 2<head> 3<meta charset=utf-8> 4<meta name=viewport content="width=device-width,initial-scale=1"> 5<meta http-equiv=x-ua-compatible content="IE=edge"> 6<meta name=generator content="Wowchemy 5.3.0 for Hugo"> 7<link rel=preconnect href=https://fonts.gstatic.com crossorigin> 8<link rel=preload as=style href="https://fonts.googleapis.com/css2?family=Fira+Code:wght@300&family=Fira+Sans:wght@200;Montserrat:wght@400;700&family=Roboto+Mono&family=Roboto:wght@400;700&display=swap"> 9<link rel=stylesheet href="https://fonts.googleapis.com/css2?family=Fira+Code:wght@300&family=Fira+Sans:wght@200;Montserrat:wght@400;700&family=Roboto+Mono&family=Roboto:wght@400;700&display=swap" media=print onload="this.media='all'"> 10<meta name=author content="Pascal Notin"> 11<meta name=description content> 12<link rel=alternate hreflang=en-us href=https://www.pascalnotin.com/publication/> 13<meta name=theme-color content="#1565c0"> 14<link rel=stylesheet href=/css/vendor-bundle.min.f1ecf783c14edc00c9320c205831ad8e.css media=print onload="this.media='all'"> 15<link rel=stylesheet href=https://cdn.jsdelivr.net/npm/[email protected]/css/academicons.min.css integrity="sha512-W0xM4mr6dEP9nREo7Z9z+9X70wytKvMGeDsj7ps2+xg5QPrEBXC8tAW1IFnzjR6eoJ90JmCnFzerQJTLzIEHjA==" crossorigin=anonymous media=print onload="this.media='all'"> 16<link rel=stylesheet href=https://cdn.jsdelivr.net/gh/highlightjs/[email protected]/build/styles/github.min.css crossorigin=anonymous title=hl-light media=print onload="this.media='all'"> 17<link rel=stylesheet href=https://cdn.jsdelivr.net/gh/highlightjs/[email protected]/build/styles/dracula.min.css crossorigin=anonymous title=hl-dark media=print onload="this.media='all'" disabled> 18<link rel=stylesheet href=/css/wowchemy.dd28815de0e283cec97459c287d9bbd4.css> 19<link rel=alternate href=/publication/index.xml type=application/rss+xml title="Pascal Notin"> 20<link rel=manifest href=/manifest.webmanifest> 21<link rel=icon type=image/png href=/media/icon_hu5202dac0a2755c9a62d3a907d75f2b8b_108322_32x32_fill_lanczos_center_3.png> 22<link rel=apple-touch-icon type=image/png href=/media/icon_hu5202dac0a2755c9a62d3a907d75f2b8b_108322_180x180_fill_lanczos_center_3.png> 23<link rel=canonical href=https://www.pascalnotin.com/publication/> 24<meta property="twitter:card" content="summary"> 25<meta property="og:site_name" content="Pascal Notin"> 26<meta property="og:url" content="https://www.pascalnotin.com/publication/"> 27<meta property="og:title" content="Selected Publications | Pascal Notin"> 28<meta property="og:description" content><meta property="og:image" content="https://www.pascalnotin.com/media/logo_hu5202dac0a2755c9a62d3a907d75f2b8b_108322_300x300_fit_lanczos_3.png"> 29<meta property="twitter:image" content="https://www.pascalnotin.com/media/logo_hu5202dac0a2755c9a62d3a907d75f2b8b_108322_300x300_fit_lanczos_3.png"><meta property="og:locale" content="en-us"> 30<link rel=stylesheet href=https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.6.0/css/all.min.css integrity="sha512-d+LpWiwKuQgPeRppp1CeBEFQ1IYeYVynoljmOiakWRUJ8xTGXmDwLJaXYIVFVWZ8MUIS1xzKHP9HoOjVgvBhQQ==" crossorigin=anonymous referrerpolicy=no-referrer>
31<script>document.addEventListener('DOMContentLoaded',function(){if(window.location.pathname!=='/'){let a=document.title;a.includes(' | ')&&(document.title=a.split(' | ')[0])}})</script>
31 32<title>Selected Publications | Pascal Notin</title> 33</head> 34<body id=top data-spy=scroll data-offset=70 data-target=#TableOfContents class=page-wrapper data-wc-page-id=3a079e7dad19be978a318345a7749d34>
35<script src=/js/wowchemy-init.min.da42964e270e05a6063fbca894d7ccff.js></script>
35 36<aside class=search-modal id=search> 37<div class=container> 38<section class=search-header> 39<div class="row no-gutters justify-content-between mb-3"> 40<div class=col-6> 41<h1>Search</h1> 42</div> 43<div class="col-6 col-search-close"> 44<a class=js-search href=# aria-label=Close><i class="fas fa-times-circle text-muted" aria-hidden=true></i></a> 45</div> 46</div> 47<div id=search-box> 48<input name=q id=search-query placeholder=Search... autocapitalize=off autocomplete=off autocorrect=off spellcheck=false type=search class=form-control aria-label=Search...> 49</div> 50</section> 51<section class=section-search-results> 52<div id=search-hits> 53</div> 54</section> 55</div> 56</aside> 57<div class=page-header> 58<nav class="navbar navbar-expand-lg navbar-light compensate-for-scrollbar" id=navbar-main> 59<div class=container-xl> 60<div class="d-none d-lg-inline-flex"> 61<a class=navbar-brand href=/><img src=/media/logo_hu5202dac0a2755c9a62d3a907d75f2b8b_108322_0x70_resize_lanczos_3.png alt="Pascal Notin"></a> 62</div> 63<button type=button class=navbar-toggler data-toggle=collapse data-target=#navbar-content aria-controls=navbar-content aria-expanded=false aria-label="Toggle navigation"> 64<span><i class="fas fa-bars"></i></span> 65</button> 66<div class="navbar-brand-mobile-wrapper d-inline-flex d-lg-none"> 67<a class=navbar-brand href=/><img src=/media/logo_hu5202dac0a2755c9a62d3a907d75f2b8b_108322_0x70_resize_lanczos_3.png alt="Pascal Notin"></a> 68</div> 69<div class="navbar-collapse main-menu-item collapse justify-content-end" id=navbar-content> 70<ul class="navbar-nav d-md-inline-flex"> 71<li class=nav-item> 72<a class=nav-link href=/><span>Home</span></a> 73</li> 74<li class=nav-item> 75<a class="nav-link active" href=/publication/><span>Publications</span></a> 76</li> 77<li class=nav-item> 78<a class=nav-link href=/about/><span>About</span></a> 79</li> 80<li class=nav-item> 81<a class=nav-link href=https://github.com/pascalnotin target=_blank rel=noopener><span>Code <i class="fab fa-github"></i></span></a> 82</li> 83<li class=nav-item> 84<a class=nav-link href=https://pascalnotin.substack.com/ target=_blank rel=noopener><span>Blog <i class="fas fa-rss"></i></span></a> 85</li> 86</ul> 87</div> 88<ul class="nav-icons navbar-nav flex-row ml-auto d-flex pl-md-2"> 89<li class=nav-item> 90<a class="nav-link js-search" href=# aria-label=Search><i class="fas fa-search" aria-hidden=true></i></a> 91</li> 92</ul> 93</div> 94</nav> 95</div> 96<div class=page-body> 97<div class="universal-wrapper pt-3"> 98<h1>Selected Publications</h1> 99</div> 100<div class=universal-wrapper> 101<div class=row> 102<div class=col-lg-12> 103<div class=article-style><p><a href="https://scholar.google.com/citations?user=soxv0s0AAAAJ&hl=en" target=_blank rel=noopener>Google scholar</a> | <a href=https://www.semanticscholar.org/author/Pascal-Notin/1825748945 target=_blank rel=noopener>Semantic scholar</a> | <a href=https://orcid.org/0000-0002-1877-8983 target=_blank rel=noopener>ORCID</a></p> 104</div> 105<div class="form-row mb-4"> 106<div class=col-auto> 107<input type=search class="filter-search form-control form-control-sm" placeholder=Search... autocapitalize=off autocomplete=off autocorrect=off role=textbox spellcheck=false> 108</div> 109<div class=col-auto> 110<select class="pub-filters pubtype-select form-control form-control-sm" data-filter-group=pubtype><option value=*>Type</option><option value=.pubtype-1> 111Conference paper 112</option><option value=.pubtype-2> 113Journal article 114</option><option value=.pubtype-3> 115Preprint 116</option></select> 117</div> 118<div class=col-auto> 119<select class="pub-filters form-control form-control-sm" data-filter-group=year><option value=*>Date</option><option value=.year-2025> 1202025 121</option><option value=.year-2024> 1222024 123</option><option value=.year-2023> 1242023 125</option><option value=.year-2022> 1262022 127</option><option value=.year-2021> 1282021 129</option></select> 130</div> 131</div> 132<div id=container-publications> 133<div class="grid-sizer col-lg-12 isotope-item pubtype-3 year-2025"> 134<div class="media stream-item view-compact"> 135<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 136<a href=/publication/rnagym/> 137<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 138<img src=/publication/rnagym/featured_hu77b024dcf09961dc7501c8f631ce1553_186949_150x100_fit_lanczos_3.png srcset="/publication/rnagym/featured_hu77b024dcf09961dc7501c8f631ce1553_186949_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="RNAGym: Large-scale Benchmarks for RNA Fitness and Structure Prediction"> 139</div> 140</a> 141</div> 142<div class=media-body> 143<h3 class="article-title mb-0 mt-0"> 144<a href=/publication/rnagym/>RNAGym: Large-scale Benchmarks for RNA Fitness and Structure Prediction</a> 145</h3> 146<div class=article-metadata> 147<span> 148Rohit Arora</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 149Murphy Angelo</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 150Christian Andrew Choe</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 151Courtney A. Shearer</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 152Aaron W. Kollasch</span>, <span> 153Fiona Qu</span>, <span> 154Ruben Weitzman</span>, <span> 155Artem Gazizov</span>, <span> 156Sarah Gurev</span>, <span> 157Erik Xie</span>, <span> 158Debora Marks</span>, <span class=author-highlighted> 159Pascal Notin</span> 160</div> 161<div class=article-style> 162<b>Preprint.</b> Large-scale benchmarks to assess models for RNA fitness & structure prediction. 163</div> 164<div class="btn-links mt-2"> 165<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.bi
165orxiv.org/content/10.1101/2025.06.16.660049v1.full.pdf target=_blank rel=noopener> 166PDF 167</a> 168<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/rnagym/cite.bib> 169Cite 170</a> 171<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/MarksLab-DasLab/RNAGym target=_blank rel=noopener> 172Code 173</a> 174<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2025.06.16.660049v1 target=_blank rel=noopener> 175bioRxiv</a> 176</div> 177</div> 178</div> 179<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 180</div> 181<div class="grid-sizer col-lg-12 isotope-item pubtype-2 year-2025"> 182<div class="media stream-item view-compact"> 183<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 184<a href=/publication/adk/> 185<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 186<img src=/publication/adk/featured_hua35911ab655f8c22a241a08c0bd042b6_1875424_150x100_fit_lanczos_3.png srcset="/publication/adk/featured_hua35911ab655f8c22a241a08c0bd042b6_1875424_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Evolutionary-Scale Enzymology Enables Biochemical Constant Prediction Across a Multi-Peaked Catalytic Landscape"> 187</div> 188</a> 189</div> 190<div class=media-body> 191<h3 class="article-title mb-0 mt-0"> 192<a href=/publication/adk/>Evolutionary-Scale Enzymology Enables Biochemical Constant Prediction Across a Multi-Peaked Catalytic Landscape</a> 193</h3> 194<div class=article-metadata> 195<span> 196Duncan F. Muir</span>, <span> 197Garrison P. R. Asper</span>, <span class=author-highlighted> 198Pascal Notin</span>, <span> 199Jacob A. Posner</span>, <span> 200Debora S. Marks</span>, <span> 201Michael J. Keiser</span>, <span> 202Margaux M. Pinney</span> 203</div> 204<div class=article-style> 205<b>Science, 2025.</b> A microfluidic platform to measure catalytic constants for hundreds of Adenylate Kinase variants, enabling mapping of the sequence-catalysis landscape and development of superior predictive models. 206</div> 207<div class="btn-links mt-2"> 208<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2024.10.23.619915v1.full.pdf target=_blank rel=noopener> 209PDF 210</a> 211<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/adk/cite.bib> 212Cite 213</a> 214<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/pinneylab/adk_evolutionary_landscape target=_blank rel=noopener> 215Code 216</a> 217<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2024.10.23.619915v1 target=_blank rel=noopener> 218arXiv</a> 219<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.science.org/doi/10.1126/science.adu1058 target=_blank rel=noopener> 220Science</a> 221</div> 222</div> 223</div> 224<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 225</div> 226<div class="grid-sizer col-lg-12 isotope-item pubtype-2 year-2025"> 227<div class="media stream-item view-compact"> 228<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 229<a href=/publication/protriever/> 230<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 231<img src=/publication/protriever/featured_huca0447e53a6531cb3122374f682b0808_811736_150x100_fit_lanczos_3.png srcset="/publication/protriever/featured_huca0447e53a6531cb3122374f682b0808_811736_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Protriever: End-to-End Differentiable Protein Homology Search for Fitness Prediction"> 232</div> 233</a> 234</div> 235<div class=media-body> 236<h3 class="article-title mb-0 mt-0"> 237<a href=/publication/protriever/>Protriever: End-to-End Differentiable Protein Homology Search for Fitness Prediction</a> 238</h3> 239<div class=article-metadata> 240<span> 241Ruben Weitzman</span>, <span> 242Peter Mørch Groth</span>, <span> 243Lood van Niekerk</span>, <span> 244Aoi Otani</span>, <span> 245Yarin Gal</span>, <span> 246Debora Marks</span>, <span class=author-highlighted> 247Pascal Notin</span> 248</div> 249<div class=article-style> 250<b>ICML, 2025.</b>
250 an end-to-end differentiable framework that learns to retrieve relevant protein homologs while simultaneously training for downstream tasks, achieving state-of-the-art fitness prediction performance while being orders of magnitude faster than traditional MSA-based approaches. 251</div> 252<div class="btn-links mt-2"> 253<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.arxiv.org/pdf/2506.08954 target=_blank rel=noopener> 254PDF 255</a> 256<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/protriever/cite.bib> 257Cite 258</a> 259<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/OATML-Markslab/Protriever target=_blank rel=noopener> 260Code 261</a> 262<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://proceedings.mlr.press/v267/weitzman25a.html target=_blank rel=noopener> 263Proceedings</a> 264<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.arxiv.org/abs/2506.08954 target=_blank rel=noopener> 265arXiv</a> 266</div> 267</div> 268</div> 269<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 270</div> 271<div class="grid-sizer col-lg-12 isotope-item pubtype-3 year-2025"> 272<div class="media stream-item view-compact"> 273<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 274<a href=/publication/cooperativity/> 275<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 276<img src=/publication/cooperativity/featured_hu41833a76f9e6ab80c076a0b1a8876629_1017523_150x100_fit_lanczos_3.png srcset="/publication/cooperativity/featured_hu41833a76f9e6ab80c076a0b1a8876629_1017523_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Large-scale discovery, analysis, and design of protein energy landscapes"> 277</div> 278</a> 279</div> 280<div class=media-body> 281<h3 class="article-title mb-0 mt-0"> 282<a href=/publication/cooperativity/>Large-scale discovery, analysis, and design of protein energy landscapes</a> 283</h3> 284<div class=article-metadata> 285<span> 286Ãllan J. R. Ferrari</span>, <span> 287Sugyan M. Dixit</span>, <span> 288Jane Thibeault</span>, <span> 289Mario Garcia</span>, <span> 290Scott Houliston</span>, <span> 291Robert W. Ludwig</span>, <span class=author-highlighted> 292Pascal Notin</span>, <span> 293Claire M. Phoumyvong</span>, <span> 294Cydney M. Martell</span>, <span> 295Michelle D. Jung</span>, <span> 296Kotaro Tsuboyama</span>, <span> 297Lauren Carter</span>, <span> 298Cheryl H. Arrowsmith</span>, <span> 299Miklos Guttman</span>, <span> 300Gabriel J. Rocklin</span> 301</div> 302<div class=article-style> 303<b>Preprint.</b> A multiplexed experimental approach to analyze conformational fluctuations across thousands of protein domains, revealing hidden variations that affect protein cooperativity and function. 304</div> 305<div class="btn-links mt-2"> 306<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2025.03.20.644235v1.full.pdf target=_blank rel=noopener> 307PDF 308</a> 309<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/cooperativity/cite.bib> 310Cite 311</a> 312<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2025.03.20.644235v1.abstract target=_blank rel=noopener> 313arXiv</a> 314</div> 315</div> 316</div> 317<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 318</div> 319<div class="grid-sizer col-lg-12 isotope-item pubtype-1 year-2024"> 320<div class="media stream-item view-compact"> 321<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 322<a href=/publication/s3f/> 323<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 324<img src=/publication/s3f/featured_hue261e40539947bc5520eb2a05de526c9_1810967_150x100_fit_lanczos_3.png srcset="/publication/s3f/featured_hue261e40539947bc5520eb2a05de526c9_1810967_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Multi-Scale Representation Learning for Protein Fitness Prediction"> 325</div> 326</a> 327</div> 328<div class=media-body> 329<h3 class="article-title mb-0 mt-0"> 330<a href=/publication/s3f/>Multi-Scale Representation Learning for Protein Fitness Prediction</a> 331</h3> 332<div class=article-metadata> 333<span> 334Zuobai Zhang</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span class=author-highlighted> 335Pascal Notin</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 336Yining Huang</span>, <span> 337Aurélie Lozano</span>, <span> 338Vijil Chenthamarakshan</span>, <span> 339Debora Marks</span>, <span> 340Payel Das</span>, <span> 341Jian Tang</span> 342</div> 343<div class=article-style> 344<b>NeurIPS, 2024.</b> A multimodal framework that integrates protein sequence, structure, and surface topology features to achieve state-of-the-art fitness prediction. 345</div> 346<div class="btn-links mt-2"> 347<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://proceedings.neurips.cc/paper_files/paper/2024/file/b7d795e655c1463d7299688d489e8ef4-Paper-Conference.pdf target=_blank rel=noopener> 348PDF 349</a> 350<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/s3f/cite.bib> 351Cite 352</a> 353<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/DeepGraphLearning/S3F target=_blank rel=noopener> 354Code 355</a> 356<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/abs/2412.01108 target=_blank rel=noopener> 357arXiv</a> 358<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://proceedings.neurips.cc/paper_files/paper/2024/hash/b7d795e655c1463d7299688d489e8ef4-Abstract-Conference.html target=_blank rel=noopener> 359Proceedings</a> 360</div> 361</div> 362</div> 363<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 364</div> 365<div class="grid-sizer col-lg-12 isotope-item pubtype-3 year-2024"> 366<div class="media stream-item view-compact"> 367<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 368<a href=/publication/lol-eve/> 369<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 370<img src=/publication/lol-eve/featured_hu1046f339ecaaacb944e926b5f50341d6_403975_150x100_fit_lanczos_3.png srcset="/publication/lol-eve/featured_hu1046f339ecaaacb944e926b5f50341d6_403975_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Predicting Promoter Variant Effe
370cts from Evolutionary Sequences"> 371</div> 372</a> 373</div> 374<div class=media-body> 375<h3 class="article-title mb-0 mt-0"> 376<a href=/publication/lol-eve/>Predicting Promoter Variant Effects from Evolutionary Sequences</a> 377</h3> 378<div class=article-metadata> 379<span> 380Courtney A. Shearer</span>, <span> 381Felix Teufel</span>, <span> 382Rose Orenbuch</span>, <span> 383Daniel Ritter</span>, <span> 384Aviv Spinner</span>, <span> 385Erik Xie</span>, <span> 386Jonathan Frazer</span>, <span> 387Mafalda Dias</span>, <span class=author-highlighted> 388Pascal Notin</span>, <span> 389Debora S. Marks</span> 390</div> 391<div class=article-style> 392<b>Preprint.</b> A conditional autoregressive transformer model trained on 14.6 million mammalian promoter sequences that achieves state-of-the-art performance in predicting the effects of indels in human promoter regions. 393</div> 394<div class="btn-links mt-2"> 395<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2024.11.11.623015v1.full.pdf target=_blank rel=noopener> 396PDF 397</a> 398<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/lol-eve/cite.bib> 399Cite 400</a> 401<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2024.11.11.623015v1 target=_blank rel=noopener> 402arXiv</a> 403</div> 404</div> 405</div> 406<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 407</div> 408<div class="grid-sizer col-lg-12 isotope-item pubtype-2 year-2024"> 409<div class="media stream-item view-compact"> 410<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 411<a href=/publication/ml_functional_protein_design/> 412<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 413<img src=/publication/ml_functional_protein_design/featured_huc72c9303025b017e68a00e09e1293e2f_303989_150x100_fit_lanczos_3.png srcset="/publication/ml_functional_protein_design/featured_huc72c9303025b017e68a00e09e1293e2f_303989_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Machine Learning for Functional Protein Design"> 414</div> 415</a> 416</div> 417<div class=media-body> 418<h3 class="article-title mb-0 mt-0"> 419<a href=/publication/ml_functional_protein_design/>Machine Learning for Functional Protein Design</a> 420</h3> 421<div class=article-metadata> 422<span class=author-highlighted> 423Pascal Notin</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 424Nathan Rollins</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 425Yarin Gal</span>, <span> 426Chris Sander</span>, <span> 427Debora Marks</span> 428</div> 429<div class=article-style> 430<b>Nature Biotech, 2024.</b> A unifying framework that classifies models on the basis of their use of three core data modalities: sequences, structures and functional labels is introduced to make sense of the exploding diversity of machine learning approaches. 431</div> 432<div class="btn-links mt-2"> 433<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/ml_functional_protein_design/cite.bib> 434Cite 435</a> 436<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.nature.com/articles/s41587-024-02127-0 target=_blank rel=noopener> 437Paper</a> 438<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/1epxpuAzPk_ZNQTkJZjPdzFAkUWkrQN0A/view?usp=sharing" target=_blank rel=noopener> 439NBT Cover</a> 440<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.nature.com/nbt/volumes/42/issues/2 target=_blank rel=noopener> 441NBT volume</a> 442</div> 443</div> 444</div> 445<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 446</div> 447<div class="grid-sizer col-lg-12 isotope-item pubtype-1 year-2023"> 448<div class="media stream-item view-compact"> 449<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 450<a href=/publication/proteingym/> 451<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 452<img src=/publication/proteingym/featured_hucca3c694c30cd94be68629813db931a3_282909_150x100_fit_lanczos_3.png srcset="/publication/proteingym/featured_hucca3c694c30cd94be68629813db931a3_282909_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="ProteinGym: Large-Scale Benchmarks for Protein Fitness Prediction and Design"> 453</div> 454</a> 455</div> 456<div class=media-body> 457<h3 class="article-title mb-0 mt-0"> 458<a href=/publication/proteingym/>ProteinGym: Large-Scale Benchmarks for Protein Fitness Prediction and Design</a> 459</h3> 460<div class=article-metadata> 461<span class=author-highlighted> 462Pascal Notin</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 463Aaron W. Kollasch</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 464Daniel Ritter</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 465Lood van Niekerk</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 466Steffanie Paul</span>, <span> 467Hansen Spinner</span>, <span> 468Nathan Rollins</span>, <span> 469Ada Y. Shaw</span>, <span> 470Ruben Weitzman</span>, <span> 471Jonathan Frazer</span>, <span> 472Mafalda Dias</span>, <span> 473Dinko Franceschi</span>, <span> 474Rose Orenbuch</span>, <span> 475Debora S. Marks</span>, <span> 476Yarin Gal</span> 477</div> 478<div class=article-style> 479<b>NeurIPS, 2023.</b> Large-scale benchmarks to assess models for protein fitness prediction and design. 480</div> 481<div class="btn-links mt-2"> 482<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://papers.nips.cc/paper_files/paper/2023/file/cac723e5ff29f65e3fcbb0739ae91bee-Paper-Datasets_and_Benchmarks.pdf target=_blank rel=noopener> 483PDF 484</a> 485<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/proteingym/cite.bib> 486Cite 487</a> 488<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/OATML-Markslab/ProteinGym target=_blank rel=noopener> 489Code 490</a> 491<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/1BvESvInK9s1MznRhKgrhG0qb8z4ST3D-/view?usp=sharing" target=_blank rel=noopener> 492Poster 493</a> 494<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/1e2ARr9VO0aOIwghzE2bh-RELPB7PMUx0/view?usp=sharing" target=_blank rel=noopener> 495Slides 496</a> 497<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://papers.nips.cc/paper_files/paper/2023/hash/cac723e5ff29f65e3fcbb0739ae91bee-Abstract-Datasets_and_Benchmarks.html target=_blank rel=noopener> 498Proceedings</a> 499<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.bi
499orxiv.org/content/10.1101/2023.12.07.570727v1 target=_blank rel=noopener> 500bioRxiv</a> 501<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://proteingym.org/ target=_blank rel=noopener> 502Website</a> 503</div> 504</div> 505</div> 506<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 507</div> 508<div class="grid-sizer col-lg-12 isotope-item pubtype-1 year-2023"> 509<div class="media stream-item view-compact"> 510<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 511<a href=/publication/proteinnpt/> 512<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 513<img src=/publication/proteinnpt/featured_hud99eb5f3dec7b4ce2ff5aa370f179af4_488213_150x100_fit_lanczos_3.png srcset="/publication/proteinnpt/featured_hud99eb5f3dec7b4ce2ff5aa370f179af4_488213_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="ProteinNPT: Improving Protein Property Prediction and Design with Non-Parametric Transformers"> 514</div> 515</a> 516</div> 517<div class=media-body> 518<h3 class="article-title mb-0 mt-0"> 519<a href=/publication/proteinnpt/>ProteinNPT: Improving Protein Property Prediction and Design with Non-Parametric Transformers</a> 520</h3> 521<div class=article-metadata> 522<span class=author-highlighted> 523Pascal Notin</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 524Ruben Weitzman</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 525Debora S. Marks</span>, <span> 526Yarin Gal</span> 527</div> 528<div class=article-style> 529<b>NeurIPS, 2023.</b> A conditional semi-supervised pseudo-generative model for fitness prediction and design. 530</div> 531<div class="btn-links mt-2"> 532<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://papers.nips.cc/paper_files/paper/2023/file/6a4d5d85f7a52f062d23d98d544a5578-Paper-Conference.pdf target=_blank rel=noopener> 533PDF 534</a> 535<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/proteinnpt/cite.bib> 536Cite 537</a> 538<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/OATML-Markslab/ProteinNPT target=_blank rel=noopener> 539Code 540</a> 541<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/1yST8xIoLz7ID_FdCFx0IJ8cxtWUciI3f/view?usp=sharing" target=_blank rel=noopener> 542Poster 543</a> 544<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/1t7rCfgvzdmThFel9mv9OZkPH4OOjsWJD/view?usp=sharing" target=_blank rel=noopener> 545Slides 546</a> 547<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://papers.nips.cc/paper_files/paper/2023/hash/6a4d5d85f7a52f062d23d98d544a5578-Abstract-Conference.html target=_blank rel=noopener> 548Proceedings</a> 549<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2023.12.06.570473v1 target=_blank rel=noopener> 550bioRxiv</a> 551</div> 552</div> 553</div> 554<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 555</div> 556<div class="grid-sizer col-lg-12 isotope-item pubtype-2 year-2023"> 557<div class="media stream-item view-compact"> 558<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 559<a href=/publication/viral_escape_journal/> 560<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 561<img src=/publication/viral_escape_journal/featured_huf18684ea3b32272ec0f64d529ab4180c_807486_150x100_fit_lanczos_3.png srcset="/publication/viral_escape_journal/featured_huf18684ea3b32272ec0f64d529ab4180c_807486_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Learning from Prepandemic Data to Forecast Viral Escape"> 562</div> 563</a> 564</div> 565<div class=media-body> 566<h3 class="article-title mb-0 mt-0"> 567<a href=/publication/viral_escape_journal/>Learning from Prepandemic Data to Forecast Viral Escape</a> 568</h3> 569<div class=article-metadata> 570<span> 571Nikki Thadani</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 572Sarah Gurev</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span class=author-highlighted> 573Pascal Notin</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 574Noor Youssef</span>, <span> 575Nathan Rollins</span>, <span> 576Daniel Ritter</span>, <span> 577Chris Sander</span>, <span> 578Yarin Gal</span>, <span> 579Debora Marks</span> 580</div> 581<div class=article-style> 582<b>Nature, 2023.</b> A computational framework to predict viral escape from pre-pandemic information only (evolutionary data and 3D structure). 583</div> 584<div class="btn-links mt-2"> 585<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/viral_escape_journal/cite.bib> 586Cite 587</a> 588<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/OATML-Markslab/EVEscape target=_blank rel=noopener> 589Code 590</a> 591<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.nature.com/articles/s41586-023-06617-0 target=_blank rel=noopener> 592Nature</a> 593<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.bi
593orxiv.org/content/10.1101/2022.07.21.501023v1 target=_blank rel=noopener> 594bioRxiv</a> 595<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.evescape.org/ target=_blank rel=noopener> 596Website</a> 597</div> 598</div> 599</div> 600<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 601</div> 602<div class="grid-sizer col-lg-12 isotope-item pubtype-1 year-2023"> 603<div class="media stream-item view-compact"> 604<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 605<a href=/publication/discobax/> 606<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 607<img src=/publication/discobax/featured_hua7e528efbe4b1b482c892ba950f0ab52_436637_150x100_fit_lanczos_3.png srcset="/publication/discobax/featured_hua7e528efbe4b1b482c892ba950f0ab52_436637_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="DiscoBAX: Discovery of Optimal Intervention Sets in Genomic Experiment Design"> 608</div> 609</a> 610</div> 611<div class=media-body> 612<h3 class="article-title mb-0 mt-0"> 613<a href=/publication/discobax/>DiscoBAX: Discovery of Optimal Intervention Sets in Genomic Experiment Design</a> 614</h3> 615<div class=article-metadata> 616<span> 617Clare Lyle</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 618Arash Mehrjou</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span class=author-highlighted> 619Pascal Notin</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 620Andrew Jesson</span>, <span> 621Stefan Bauer</span>, <span> 622Yarin Gal</span>, <span>
623Patrick Schwab</span> 624</div> 625<div class=article-style> 626<b>ICML, 2023.</b> A sample-efficient method for discovering optimal sets that are both diverse and optimize the function of interest. 627</div> 628<div class="btn-links mt-2"> 629<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/discobax/cite.bib> 630Cite 631</a> 632<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/amehrjou/discobax target=_blank rel=noopener> 633Code 634</a> 635<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/abs/2312.04064 target=_blank rel=noopener> 636arXiv</a> 637<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://proceedings.mlr.press/v202/lyle23a.html target=_blank rel=noopener> 638Proceedings</a> 639<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.gsk.ai/blogs/discobax-discovery-of-optimal-intervention-sets-in-genomic-experiment-design/ target=_blank rel=noopener> 640Blog</a> 641</div> 642</div> 643</div> 644<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 645</div> 646<div class="grid-sizer col-lg-12 isotope-item pubtype-3 year-2022"> 647<div class="media stream-item view-compact"> 648<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 649<a href=/publication/trancepteve/> 650<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 651<img src=/publication/trancepteve/featured_hu7c0b16f2dd3d217b88ed5d8c00b18a6a_891600_150x100_fit_lanczos_3.png srcset="/publication/trancepteve/featured_hu7c0b16f2dd3d217b88ed5d8c00b18a6a_891600_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="TranceptEVE: Combining Family-specific and Family-agnostic Models of Protein Sequences for Improved Fitness Prediction"> 652</div> 653</a> 654</div> 655<div class=media-body> 656<h3 class="article-title mb-0 mt-0"> 657<a href=/publication/trancepteve/>TranceptEVE: Combining Family-specific and Family-agnostic Models of Protein Sequences for Improved Fitness Prediction</a> 658</h3> 659<div class=article-metadata> 660<span class=author-highlighted> 661Pascal Notin</span>, <span> 662Lood van Niekerk</span>, <span> 663Aaron W Kollasch</span>, <span> 664Daniel Ritter</span>, <span> 665Yarin Gal</span>, <span> 666Debora S. Marks</span> 667</div> 668<div class=article-style> 669<b>NeurIPS, LMRL, 2022.</b> A hybrid family-specific and family-agnostic model to achieve SOTA performance on protein fitness prediction and human variant annotation. 670</div> 671<div class="btn-links mt-2"> 672<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/trancepteve/cite.bib> 673Cite 674</a> 675<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/12Xh_kWXz8i-kP1qpfuKOxoTmSra6VRf9/view?usp=sharing" target=_blank rel=noopener> 676Slides 677</a> 678<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://www.youtube.com/watch?v=m0QVNWcRi8Y" target=_blank rel=noopener> 679Video 680</a> 681<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2022.12.07.519495v2 target=_blank rel=noopener> 682bioRxiv</a> 683</div> 684</div> 685</div> 686<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 687</div> 688<div class="grid-sizer col-lg-12 isotope-item pubtype-1 year-2022"> 689<div class="media stream-item view-compact"> 690<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 691<a href=/publication/tranception/> 692<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 693<img src=/publication/tranception/featured_hu77345be8db48aee35ad7fe2b4a6f449c_628059_150x100_fit_lanczos_3.png srcset="/publication/tranception/featured_hu77345be8db48aee35ad7fe2b4a6f449c_628059_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Tranception: Protein Fitness Prediction with Autoregressive Transformers and Inference-time Retrieval"> 694</div> 695</a> 696</div> 697<div class=media-body> 698<h3 class="article-title mb-0 mt-0"> 699<a href=/publication/tranception/>Tranception: Protein Fitness Prediction with Autoregressive Transformers and Inference-time Retrieval</a> 700</h3> 701<div class=article-metadata> 702<span class=author-highlighted> 703Pascal Notin</span>, <span> 704Mafalda Dias</span>, <span> 705Jonathan Frazer</span>, <span> 706Javier Marchena-Hurtado</span>, <span>
707Aidan N. Gomez</span>, <span> 708Debora S. Marks</span>, <span> 709Yarin Gal</span> 710</div> 711<div class=article-style> 712<b>ICML, 2022.</b> A suite of autoregressive transformers with biological priors, augmented with inference-time retrieval, to achieve SOTA performance on protein fitness prediction. 713</div> 714<div class="btn-links mt-2"> 715<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://proceedings.mlr.press/v162/notin22a/notin22a.pdf target=_blank rel=noopener> 716PDF 717</a> 718<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/tranception/cite.bib> 719Cite 720</a> 721<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/OATML-Markslab/Tranception target=_blank rel=noopener> 722Code 723</a> 724<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/1G7psWPM21kVPsZhR-9wlg10HM4jJvNaI/view?usp=sharing" target=_blank rel=noopener> 725Poster 726</a> 727<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/16zB2gAZjpYFM2KKhcb4fZFpZXenHuPwY/view?usp=sharing" target=_blank rel=noopener> 728Slides 729</a> 730<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://slideslive.com/38983471 target=_blank rel=noopener> 731Video 732</a> 733<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/abs/2205.13760 target=_blank rel=noopener> 734arXiv</a> 735<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://proceedings.mlr.press/v162/notin22a.html target=_blank rel=noopener> 736Proceedings</a> 737<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://huggingface.co/OATML-Markslab/Tranception_Large target=_blank rel=noopener> 738HF Hub</a> 739<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.proteingym.org/ target=_blank rel=noopener> 740ProteinGym</a> 741<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://huggingface.co/spaces/PascalNotin/Tranception_design target=_blank rel=noopener> 742Design app (HF)</a> 743<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://colab.research.google.com/drive/12ni4U1Na9VWwnwdpGzYHCkNC7sud9SLy?usp=sharing" target=_blank rel=noopener> 744Design app (Colab)</a> 745</div> 746</div> 747</div> 748<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 749</div> 750<div class="grid-sizer col-lg-12 isotope-item pubtype-3 year-2022"> 751<div class="media stream-item view-compact"> 752<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 753<a href=/publication/rita/> 754<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 755<img src=/publication/rita/featured_hub6bab710b396b38cf7010a029c541554_57648_150x100_fit_q100_lanczos.jpeg srcset="/publication/rita/featured_hub6bab710b396b38cf7010a029c541554_57648_300x200_fit_q100_lanczos.jpeg 2x" width=150 height=100 alt="RITA: a Study on Scaling Up Generative Protein Sequence Models"> 756</div> 757</a> 758</div> 759<div class=media-body> 760<h3 class="article-title mb-0 mt-0"> 761<a href=/publication/rita/>RITA: a Study on Scaling Up Generative Protein Sequence Models</a> 762</h3> 763<div class=article-metadata> 764<span> 765Daniel Hesslow</span>, <span> 766Niccoló Zanichelli</span>, <span class=author-highlighted> 767Pascal Notin</span>, <span> 768Iacopo Poli</span>, <span> 769Debora Marks</span> 770</div> 771<div class=article-style> 772<b>ICML, WCB, 2022.</b> The first paper investigating scaling laws in protein language modeling. 773</div> 774<div class="btn-links mt-2"> 775<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/pdf/2205.05789.pdf target=_blank rel=noopener> 776PDF 777</a> 778<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/rita/cite.bib> 779Cite 780</a> 781<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/lightonai/RITA target=_blank rel=noopener> 782Code 783</a> 784<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/abs/2205.05789 target=_blank rel=noopener> 785arXiv</a> 786<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://icml-compbio.github.io/ target=_blank rel=noopener> 787Workshop</a> 788</div> 789</div> 790</div> 791<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 792</div> 793<div class="grid-sizer col-lg-12 isotope-item pubtype-2 year-2021"> 794<div class="media stream-item view-compact"> 795<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 796<a href=/publication/eve/> 797<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 798<img src=/publication/eve/featured_hu46d7d37e64de569412c2a83bf7b27b11_209333_150x100_fit_lanczos_3.png srcset="/publication/eve/featured_hu46d7d37e64de569412c2a83bf7b27b11_209333_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Disease Variant Prediction with Deep Generative Models of Evolutionary Data"> 799</div> 800</a> 801</div> 802<div class=media-body> 803<h3 class="article-title mb-0 mt-0"> 804<a href=/publication/eve/>Disease Variant Prediction with Deep Generative Models of Evolutionary Data</a> 805</h3> 806<div class=article-metadata> 807<span> 808Jonathan Frazer</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span class=author-highlighted> 809Pascal Notin</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span> 810Mafalda Dias</span><i class="author-notes fas fa-info-circle" data-toggle=tooltip title="Equal contribution"></i>, <span>
811Aidan Gomez</span>, <span> 812Joseph Min</span>, <span> 813Kelly Brock</span>, <span> 814Yarin Gal</span>, <span> 815Debora Marks</span> 816</div> 817<div class=article-style> 818<b>Nature, 2021.</b> Deep Generative Models (Bayesian VAEs) of evolutionary sequences to predict the effects of missense mutations in human proteins. 819</div> 820<div class="btn-links mt-2"> 821<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/eve/cite.bib> 822Cite 823</a> 824<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/OATML-Markslab/EVE target=_blank rel=noopener> 825Code 826</a> 827<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://www.youtube.com/watch?v=ZFzhFMl30oE" target=_blank rel=noopener> 828Video 829</a> 830<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.nature.com/articles/s41586-021-04043-8 target=_blank rel=noopener> 831Nature</a> 832<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.biorxiv.org/content/10.1101/2020.12.21.423785v1 target=_blank rel=noopener> 833bioRxiv</a> 834<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.nature.com/articles/s41587-021-01187-w target=_blank rel=noopener> 835NBT review</a> 836<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://hms.harvard.edu/news/all-about-eve target=_blank rel=noopener> 837Press (HMS)</a> 838<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.thecrimson.com/article/2021/10/29/eve-artificial-intelligence-diagnosis-tool/ target=_blank rel=noopener> 839Press (Crimson)</a> 840<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.genengnews.com/news/ai-tool-eve-accurately-predicts-disease-relevance-of-human-genetic-variants/ target=_blank rel=noopener> 841Press (GEN)</a> 842<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://evemodel.org/ target=_blank rel=noopener> 843Website</a> 844</div> 845</div> 846</div> 847<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 848</div> 849<div class="grid-sizer col-lg-12 isotope-item pubtype-1 year-2021"> 850<div class="media stream-item view-compact"> 851<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 852<a href=/publication/genedisco/> 853<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 854<img src=/publication/genedisco/featured_hubb883e983d12d2157e0964b8e4788135_199041_150x100_fit_lanczos_3.png srcset="/publication/genedisco/featured_hubb883e983d12d2157e0964b8e4788135_199041_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="GeneDisco: A Benchmark for Experimental Design in Drug Discovery"> 855</div> 856</a> 857</div> 858<div class=media-body> 859<h3 class="article-title mb-0 mt-0"> 860<a href=/publication/genedisco/>GeneDisco: A Benchmark for Experimental Design in Drug Discovery</a> 861</h3> 862<div class=article-metadata> 863<span> 864Arash Mehrjou</span>, <span> 865Ashkan Soleymani</span>, <span> 866Andrew Jesson</span>, <span class=author-highlighted> 867Pascal Notin</span>, <span> 868Yarin Gal</span>, <span> 869Stefan Bauer</span>, <span>
870Patrick Schwab</span> 871</div> 872<div class=article-style> 873<b>ICLR, 2022.</b> A benchmark suite for evaluating active learning algorithms for experimental design in drug discovery. 874</div> 875<div class="btn-links mt-2"> 876<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/pdf/2110.11875.pdf target=_blank rel=noopener> 877PDF 878</a> 879<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/genedisco/cite.bib> 880Cite 881</a> 882<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/abs/2110.11875 target=_blank rel=noopener> 883arXiv</a> 884<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://www.gsk.ai/genedisco-challenge/ target=_blank rel=noopener> 885Challenge</a> 886</div> 887</div> 888</div> 889<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 890</div> 891<div class="grid-sizer col-lg-12 isotope-item pubtype-1 year-2021"> 892<div class="media stream-item view-compact"> 893<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 894<a href=/publication/uncertainty_guided_search/> 895<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 896<img src=/publication/uncertainty_guided_search/featured_hud06eae374232507f1b372ae2b9979de1_358285_150x100_fit_lanczos_3.png srcset="/publication/uncertainty_guided_search/featured_hud06eae374232507f1b372ae2b9979de1_358285_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Improving Black-box Optimization in VAE Latent Space Using Decoder Uncertainty"> 897</div> 898</a> 899</div> 900<div class=media-body> 901<h3 class="article-title mb-0 mt-0"> 902<a href=/publication/uncertainty_guided_search/>Improving Black-box Optimization in VAE Latent Space Using Decoder Uncertainty</a> 903</h3> 904<div class=article-metadata> 905<span class=author-highlighted> 906Pascal Notin</span>, <span> 907José Miguel Hernández-Lobato</span>, <span> 908Yarin Gal</span> 909</div> 910<div class=article-style> 911<b>NeurIPS, 2021.</b> A framework that uses the epistemic uncertainty of the decoder of a VAE to guide the optimization of properties of high-dimensional structured objects (e.g., molecules) in latent space. 912</div> 913<div class="btn-links mt-2"> 914<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/pdf/2107.00096.pdf target=_blank rel=noopener> 915PDF 916</a> 917<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/uncertainty_guided_search/cite.bib> 918Cite 919</a> 920<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://github.com/pascalnotin/uncertainty_guided_optimization target=_blank rel=noopener> 921Code 922</a> 923<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/1rAUO4tCXWqf15Ub_baHxqf6mdE14QhLl/view?usp=sharing" target=_blank rel=noopener> 924Slides 925</a> 926<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/abs/2107.00096 target=_blank rel=noopener> 927arXiv</a> 928<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://nips.cc/virtual/2021/poster/26287 target=_blank rel=noopener> 929Video</a> 930<a class="btn btn-outline-primary btn-page-header btn-sm" href="https://drive.google.com/file/d/1s1JPV8X14AP7fSSh5OOTR9UOHC4-tzSR/view?usp=sharing" target=_blank rel=noopener> 931Poster</a> 932<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://papers.nips.cc/paper/2021/hash/06fe1c234519f6812fc4c1baae25d6af-Abstract.html target=_blank rel=noopener> 933Proceedings</a> 934</div> 935</div> 936</div> 937<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 938</div> 939<div class="grid-sizer col-lg-12 isotope-item pubtype-3 year-2021"> 940<div class="media stream-item view-compact"> 941<div style=margin-right:20px;min-width:150px;width:150px;display:flex;align-items:center;justify-content:center> 942<a href=/publication/sliceout/> 943<div style=width:150px;height:100px;overflow:hidden;border-radius:4px> 944<img src=/publication/sliceout/featured_hu9cea37c7cd7f8fa7f8435cdb8d3370ff_704821_150x100_fit_lanczos_3.png srcset="/publication/sliceout/featured_hu9cea37c7cd7f8fa7f8435cdb8d3370ff_704821_300x200_fit_lanczos_3.png 2x" width=150 height=100 alt="Improving Compute Efficacy Frontiers with SliceOut"> 945</div> 946</a> 947</div> 948<div class=media-body> 949<h3 class="article-title mb-0 mt-0">
950<a href=/publication/sliceout/>Improving Compute Efficacy Frontiers with SliceOut</a> 951</h3> 952<div class=article-metadata> 953<span class=author-highlighted> 954Pascal Notin</span>, <span> 955Aidan N. Gomez</span>, <span> 956Joanna Yoo</span>, <span> 957Yarin Gal</span> 958</div> 959<div class=article-style> 960<b>Preprint.</b> A memory-efficient dropout-inspired scheme to train large neural networks faster with no loss in accuracy. 961</div> 962<div class="btn-links mt-2"> 963<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/pdf/2007.10909.pdf target=_blank rel=noopener> 964PDF 965</a> 966<a href=# class="btn btn-outline-primary btn-page-header btn-sm js-cite-modal" data-filename=/publication/sliceout/cite.bib> 967Cite 968</a> 969<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://arxiv.org/abs/2007.10909 target=_blank rel=noopener> 970arXiv</a> 971<a class="btn btn-outline-primary btn-page-header btn-sm" href=https://read.deeplearning.ai/the-batch/dropout-with-a-difference/ target=_blank rel=noopener> 972Press (The Batch)</a> 973</div> 974</div> 975</div> 976<hr style=margin-top:.7rem;margin-bottom:.7rem;width:100%> 977</div> 978</div> 979</div> 980</div> 981</div> 982</div> 983<div class=page-footer> 984<div class=container> 985<footer class=site-footer> 986<p class=powered-by> 987© 2021, Pascal Notin 988</p> 989<p class=powered-by> 990Published with <a href="https://wowchemy.com/?utm_campaign=poweredby" target=_blank rel=noopener>Wowchemy</a> â the free, <a href=https://github.com/wowchemy/wowchemy-hugo-themes target=_blank rel=noopener>open source</a> website builder that empowers creators. 991</p> 992</footer> 993</div> 994</div> 995<div id=modal class="modal fade" role=dialog> 996<div class=modal-dialog> 997<div class=modal-content> 998<div class=modal-header> 999<h5 class=modal-title>Cite</h5> 1000<button type=button class=close data-dismiss=modal aria-label=Close> 1001<span aria-hidden=true>×</span> 1002</button> 1003</div> 1004<div class=modal-body> 1005<pre><code class="tex hljs"></code></pre> 1006</div> 1007<div class=modal-footer> 1008<a class="btn btn-outline-primary my-1 js-copy-cite" href=# target=_blank> 1009<i class="fas fa-copy"></i> Copy 1010</a> 1011<a class="btn btn-outline-primary my-1 js-download-cite" href=# target=_blank> 1012<i class="fas fa-download"></i> Download 1013</a> 1014<div id=modal-error></div> 1015</div> 1016</div> 1017</div> 1018</div>
1019<script src=/js/vendor-bundle.min.b73dfaac3b6499dc997741748a7c3fe2.js></script>
vendor: 1 bytes, line 1019
1019
1020<script src=https://cdn.jsdelivr.net/gh/desandro/[email protected]/imagesloaded.pkgd.min.js integrity="sha512-S5PZ9GxJZO16tT9r3WJp/Safn31eu8uWrzglMahDT4dsmgqWonRY9grk3j+3tfuPr9WJNsfooOR7Gi7HL5W2jw==" crossorigin=anonymous></script>
vendor: 1 bytes, line 1020
1020
1021<script src=https://cdn.jsdelivr.net/gh/metafizzy/[email protected]/dist/isotope.pkgd.min.js integrity="sha512-Zq2BOxyhvnRFXu0+WE6ojpZLOU2jdnqbrM1hmVdGzyeCa1DgM3X5Q4A/Is9xA1IkbUeDd7755dNNI/PzSf2Pew==" crossorigin=anonymous></script>
vendor: 1 bytes, line 1021
1021
1022<script src=https://cdn.jsdelivr.net/gh/highlightjs/[email protected]/build/highlight.min.js integrity="sha512-Ypjm0o7jOxAd4hpdoppSEN0TQOC19UtPAqD+4s5AlXmUvbmmS/YMxYqAqarQYyxTnB6/rqip9qcxlNB/3U9Wdg==" crossorigin=anonymous></script>
vendor: 1 bytes, line 1022
1022
1023<script src=https://cdn.jsdelivr.net/gh/highlightjs/[email protected]/build/languages/r.min.js crossorigin=anonymous></script>
vendor: 1 bytes, line 1023
1023
1024<script id=search-hit-fuse-template type=text/x-template> 1025 <div class="search-hit" id="summary-{{key}}"> 1026 <div class="search-hit-content"> 1027 <div class="search-hit-name"> 1028 <a href="{{relpermalink}}">{{title}}</a> 1029 <div class="article-metadata search-hit-type">{{type}}</div> 1030 <p class="search-hit-description">{{snippet}}</p> 1031 </div> 1032 </div> 1033 </div> 1034 </script>
vendor: 1 bytes, line 1034
1034
1035<script src=https://cdn.jsdelivr.net/gh/krisk/[email protected]/dist/fuse.min.js integrity="sha512-o38bmzBGX+hD3JHWUFCDA09btWaqrNmoJ3RXLlrysA7PP01Kgs4UlE4MhelE1v5dJR3+cxlR4qQlotsW7jKsnw==" crossorigin=anonymous></script>
vendor: 1 bytes, line 1035
1035
1036<script src=https://cdn.jsdelivr.net/gh/julmot/[email protected]/dist/jquery.mark.min.js integrity="sha512-mhbv5DqBMgrWL+32MmsDOt/OAvqr/cHimk6B8y/bx/xS88MVkYGPiVv2ixKVrkywF2qHplNRUvFsAHUdxZ3Krg==" crossorigin=anonymous></script>
vendor: 1 bytes, line 1036
1036
1037<script src=/en/js/wowchemy.min.b050c36b66fe7ad351068b8eea2e5203.js></script>
1037 1038</body> 1039</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.