1<!DOCTYPE html> 2<html lang="" xml:lang=""> 3 <head> 4 <title>An Open Science Approach to Machine Learning in Biomedical Research</title> 5 <meta charset="utf-8" /> 6 <meta name="author" content="Batool Almarzouq" /> 7 <meta name="date" content="2021-04-08" /> 8
8<script src="libs/header-attrs/header-attrs.js"></script>
8 9 <link href="libs/remark-css/default.css" rel="stylesheet" /> 10 <link href="libs/remark-css/ninjutsu.css" rel="stylesheet" /> 11
11<script src="libs/clipboard/clipboard.min.js"></script>
11 12 <link href="libs/shareon/shareon.min.css" rel="stylesheet" /> 13
13<script src="libs/shareon/shareon.min.js"></script>
13 14 <link href="libs/xaringanExtra-shareagain/shareagain.css" rel="stylesheet" /> 15
15<script src="libs/xaringanExtra-shareagain/shareagain.js"></script>
15 16 <link rel="stylesheet" href="css/grayscale.css" type="text/css" /> 17 <link rel="stylesheet" href="css/font.css" type="text/css" /> 18 </head> 19 <body> 20 <textarea id="source"> 21 22 23 24 25 26 27class: title-slide, center, top 28background-image: url(figs/landing2.png) 29background-size: contain 30background-position: bottom 31background-size: 90% 32 33 34#### An Open Science Approach to Machine Learning in Biomedical Research 35 36#### Batool Almarzouq- @batool664 37 38--- 39 40class: left, inverse 41 42# A little bit about me! 43 44- A computational biologist affiliated with the University of Liverpool. 45- Founder of RLadies Chapter in Saudi Arabia (Dammam). 46- A curator in the R Weekly team. 47- Member of MiR accessibility committee. 48- Member in the turing way community. 49- Working on establishing an Open Science community in Saudi Arabia. 50 51# Acknowledgment 52 53- Anelda Van der 54- Malvika Sharan, Kirstie Whitaker and Martina G. Vilas 55- The Turing Way Community 56- Alison Presmanes Hill (slides) 57 58--- 59 60class: inverse, center, middle 61 62# Why do we use ML in Biomedical Research? 63 64--- 65 66background-image: url(figs/dna-to-protein2.jpg) 67background-size: contain 68background-position: 50% 1% 69background-size: 80% 70background-color: #ffffff 71 72.footnote[Image Credit: [ABC Science](https://astrobiochem.wordpress.com/grade-10-biology/chapter-10-dna-rna-and-protein-synthesis-3/)] 73-- 74 75class: bottom 76.pull-left[ 77### DNA 78- DNA sequence alignment 79- DNA sequence classification 80- DNA sequence clustering 81- DNA pattern mining 82 83Algorithms includes fuzzy sets, neural networks, genetic algorithms.] 84 85-- 86 87.pull-right[ 88 89### RNA 90- Mainly RNA-sequencing (RNA-seq) 91- Differentially expressed genes (DEGs) 92- Alternative splicing 93- Small RNA expression 94 95Algorithms include Logistic Regression, Random Forest, LMT, Random Subspace.] 96 97--- 98 99class: inverse, center, middle 100 101# Solving the sequence is not enough! 102We need to know the structure and function of the protein! 103 104--- 105 106class: center, middle 107 108background-image: url(figs/Sequence-to-function-relationship-A-Structure-function.png) 109background-size: contain 110background-position: 50% 20% 111background-size: 80% 112background-color: #ffffff 113.left[ 114.footnote[Image Credit: [doi:10.1021/cr400525m](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4095912/)]] 115-- 116 117class: center, bottom 118.bottom[ 119### How can we predict function from structure? 120To predict the function from the structure, scientists use different approaches including machine learning (ML) and deep learning algorithms 121.] 122--- 123 124class: center 125 126background-image: url(figs/SARS-CoV-2-illo-scaled.jpeg) 127background-size: contain 128background-position: 50% 40% 129background-size: 70% 130background-color: #ffffff 131.left[ 132.footnote[Credit: Supriyo Bhattacharya/Beckman Research Institute at City of Hope]] 133-- 134 135class: left, bottom 136 137.pull-bottom[ 138Prediction of protein structure is important to develop small molecules and targeted therapy for diseases. ] 139. 140--- 141 142class: inverse, center, middle 143## Why not only rely on Experemtal Methods? 144 145--- 146 147class: center, middle 148 149background-image: url(figs/growthOfDatabases.png) 150background-size: contain 151background-position: 50% 10% 152background-size: 50% 153background-color: #ffffff 154 155.left[ 156.footnote[Credit: Data for UniProtKB obtained form Claire O'Donovan via EBI database support]] 157-- 158 159class: bottom 160 161.bottom[ 162Because of the growing gap between the newly-sequenced and characterized sequences in the genome databases, computational methods in gene functional annotation are indispensable. Moreover, given the drop in the genome sequencing techniques' cost, this gap is only destined to grow.] 163. 164--- 165 166class: inverse, center, middle 167 168## Biology has become a highly data-intensive science, dependent on complex, computational, and statistical methods! 169 170--- 171 172class: inverse, center, middle 173 174## So, how can we make these methods available and accessible for researchers, while ensuring that scientific results remain reproducible? 175 176--- 177 178class: inverse, center, middle 179 180## What is the percentage of reproducible research? 181 182--- 183 184class: center, middle 185 186background-image: url(figs/reproducibility-circle2.png) 187background-size: 90% 188background-color: #f3f3f3 189 190.footnote[Credit: Key results of the survey on reproducibility conducted by Nature in 2016] 191 192--- 193 194class: inverse, center, middle 195 196# How can we overcome the reproducibility crisis? 197 198--- 199 200class: inverse, center, middle 201 202# How can you improve the reproducibility of your data science project? 203 204-- 205 206### OPEN SOURCE SOFTWARE 207 208-- 209 210### SHARE CODE/ANALYSIS 211 212-- 213 214### Share Computational ENVIRONMENT 215 216-- 217 218### VERSION CONTROL 219 220-- 221 222### TESTING 223 224-- 225 226### DOCUMENTATION 227 228-- 229 230### OPEN DATA/FAIR DATA 231 232-- 233 234### OPEN ACCESS 235 236--- 237 238class: inverse, center, middle 239 240# This is called Open Science. 241## Open Science is about extending the principles of openness to the whole research cycle, fostering sharing and collaboration as early as possible thus entailing a systemic change to the way science and research is done 242 243 -- [FOSTER Plus](https://www.fosteropenscience.eu/content/what-open-science-introduction) 244--- 245 246class: inverse, center, middle 247 248# What are the FAIR principles? 249 250--- 251 252class: center, bottom 253background-image: url(figs/Fair-principles.jpg) 254background-size: contain 255background-size: 70% 256background-color: #ffffff 257 258class: left, bottom 259.footnote[.red.bold[*] The Turing Way project illustration by Scriberia. Zenodo. http://doi.org/10.5281/zenodo.3332807] 260 261--- 262 263class: inverse, center, middle 264 265# Why do we use version control (git)? 266 267--- 268 269#### Version Control in the Old Days .. 270 271.pull-left[ 272<img src="figs/version-control1.png" width = 60%>] 273.pull-right[ 274<img src="figs/version-control2.jpeg" width = 75%> 275] 276 277--- 278 279#### Real Version Control (including backup) 280 281<img src="figs/github-vc.jpg" width = 99% center> 282 283--- 284 285class: center, middle, inverse 286 287## In the pandemic, some publishers have âopenedâ their journals to make certain articles freely available. 288 289-- 290 291## Databases have been created that are completely open access, such as the Open COVID Pledge. 292 293--- 294 295class: center, middle, inverse 296 297## UNESCO is launching international consultations aimed at developing a Recommendation on Open Science for adoption by member states in 2021 298 299-- 300 301## There is a network of Open Science Communities in Netherlands, Sweden, Germany, UK and others 302 303--- 304 305class: center, left, inverse 306 307## In line with vision 2030, we are starting an Open Science Community in Saudi Arabia. 308## It's created and developed with the help of the "Open Life Sciences" 309<img src="figs/open-lif-sci-logo.jpeg"> 310 311Open Life Sciences (OLS3) program helps individuals and stakeholders in research to become Open Science ambassadors. 312 313--- 314class: center, left, inverse 315 316We want to provide a place where newcomers and experienced peers interact, inspire each other to embed open science (research) practices and values in their workflows and provide feedback on policies, infrastru
316ctures and support services. Together working to make Open Science the norm. So we are calling out to researchers and colleagues in Saudi Arabia. 317 318.column[ 319<img src="figs/profile.jpeg" width= 95%> 320Batool Almarzouq 321The University of Liverpool 322] 323.column[ 324<img src="figs/anelda.jpeg" width= 95%> 325Founder and director of Talarify, Mentor OLS3 326] 327.column[ 328<img src="figs/paula-moraga.jpeg" width= 95%> 329Paula Moraga, Assistant Professor in Statistics for Public Health 330(KAUST) 331] 332.column[ 333 334 335---- 336 337Join me on the 24th of Feb for a workshop titled "Collaborating on Open Data Science Projects" as part of the Datathon for WiDS2021. 338] 339 340--- 341 342class: center 343#### How can you start learning about Open Science? 344 345-- 346 347background-image: url(figs/welcome-turing-way.jpg) 348background-size: contain 349background-size: 60% 350background-color: #ffffff 351 352.left[.footnote[.red.bold[*] The Turing Way project illustration by Scriberia. Zenodo. http://doi.org/10.5281/zenodo.3332807]] 353 354--- 355 356class: center, middle, inverse 357.pull-left[ 358## Kirstie Whitaker, Project Lead 359<img src="figs/kristie.jpg" width= 70%> 360] 361.pull-right[ 362## Malvika Sharan, Community Manager 363<img src="figs/malvika.jpg" width= 80%> 364] 365 366--- 367 368class: center, middle, inverse 369# So, What is the turing way? 370 371--- 372background-color: #ffffff 373class: center, middle 374<img src="figs/turing-way-com1.png" width= 300%> 375 376--- 377background-color: #404040 378<img src="figs/t4.png" width= 300%> 379 380--- 381 382background-color: #2b9686 383<img src="figs/t2.jpg" width= 100%> 384 385--- 386 387background-color: #373737 388<img src="figs/t6.png" width= 270%> 389 390--- 391 392class: center, middle, inverse 393## Join the next book dash event! 394### Book Dash November 2020 395<img src="figs/book-dash.png" width= 50%> 396 397--- 398 399class: center, middle 400## Review README.md Arabic translation 401<img src="figs/arabic.jpg" width= 90%> 402 403--- 404 405class: center, middle, inverse 406<img src="figs/t1.jpg" width= 140%> 407 408--- 409 410class: center, middle, inverse 411## Upcoming Workshop by the turing way 412Register for the free workshop '[Boost your research reproducibility with Binder](https://www.software.ac.uk/news/spaces-available-turing-way-workshop-boost-your-research-reproducibility-binder)' run by Sarah Gibson from the Turing Way as part of our Research Software Camp on research accessibility. 413 414<img src="figs/sarah.png"> 415 416--- 417 418class: inverse 419 420## Resources: 421- [The Turing Way](https://the-turing-way.netlify.app) 422 423- [The CMU ML Blog](https://blog.ml.cmu.edu/2020/08/31/5-reproducibility/) 424 425- [Redesign open science for Asia, Africa and Latin America](https://www.nature.com/articles/d41586-020-03052-3) 426 427- [Open Science Beyond Open Access: For and with communities, A step towards the decolonization of knowledge 428](https://zenodo.org/record/3946773#.YCzVmBMzY1I) 429 430- [Embracing science as it is: beyond Nobel-like research](https://www.youtube.com/watch?v=YTwM10Qob5k) 431 432- [Review on the Application of Machine Learning Algorithms in the Sequence Data Mining of DNA](https://www.frontiersin.org/articles/10.3389/fbioe.2020.01032/full) 433 434--- 435 436class: center, middle, inverse 437 438# Thank you so much! 439 440## [email protected] 441 442## Twitter: @batool664 443 444## Join RLadiesDammam: @RLadiesDammam 445 446 </textarea> 447<style data-target="print-only">@media screen {.remark-slide-container{display:block;}.remark-slide-scaler{box-shadow:none;}}</style>
448<script src="https://remarkjs.com/downloads/remark-latest.min.js"></script>
vendor: 1 bytes, line 448
448
449<script>var slideshow = remark.create({ 450"slideNumberFormat": "<div class=\"progress-bar-container\">\n <div class=\"progress-bar\" style=\"width: calc(%current% / %total% * 100%);\">\n </div>\n</div>\n", 451"highlightStyle": "solarized-light", 452"highlightLanguage": ["r", "css", "yaml"], 453"highlightLines": true, 454"countIncrementalSlides": false, 455"ratio": "16:9" 456}); 457if (window.HTMLWidgets) slideshow.on('afterShowSlide', function (slide) { 458 window.dispatchEvent(new Event('resize')); 459}); 460(function(d) { 461 var s = d.createElement("style"), r = d.querySelector(".remark-slide-scaler"); 462 if (!r) return; 463 s.type = "text/css"; s.innerHTML = "@page {size: " + r.style.width + " " + r.style.height +"; }"; 464 d.head.appendChild(s); 465})(document); 466 467(function(d) { 468 var el = d.getElementsByClassName("remark-slides-area"); 469 if (!el) return; 470 var slide, slides = slideshow.getSlides(), els = el[0].children; 471 for (var i = 1; i < slides.length; i++) { 472 slide = slides[i]; 473 if (slide.properties.continued === "true" || slide.properties.count === "false") { 474 els[i - 1].className += ' has-continuation'; 475 } 476 } 477 var s = d.createElement("style"); 478 s.type = "text/css"; s.innerHTML = "@media print { .has-continuation { display: none; } }"; 479 d.head.appendChild(s); 480})(document); 481// delete the temporary CSS (for displaying all slides initially) when the user 482// starts to view slides 483(function() { 484 var deleted = false; 485 slideshow.on('beforeShowSlide', function(slide) { 486 if (deleted) return; 487 var sheets = document.styleSheets, node; 488 for (var i = 0; i < sheets.length; i++) { 489 node = sheets[i].ownerNode; 490 if (node.dataset["target"] !== "print-only") continue; 491 node.parentNode.removeChild(node); 492 } 493 deleted = true; 494 }); 495})(); 496(function() { 497 "use strict" 498 // Replace
498<script> tags in slides area to make them executable 499 var scripts = document.querySelectorAll( 500 '.remark-slides-area .remark-slide-container script' 501 ); 502 if (!scripts.length) return; 503 for (var i = 0; i < scripts.length; i++) { 504 var s = document.createElement('script'); 505 var code = document.createTextNode(scripts[i].textContent); 506 s.appendChild(code); 507 var scriptAttrs = scripts[i].attributes; 508 for (var j = 0; j < scriptAttrs.length; j++) { 509 s.setAttribute(scriptAttrs[j].name, scriptAttrs[j].value); 510 } 511 scripts[i].parentElement.replaceChild(s, scripts[i]); 512 } 513})(); 514(function() { 515 var links = document.getElementsByTagName('a'); 516 for (var i = 0; i < links.length; i++) { 517 if (/^(https?:)?\/\//.test(links[i].getAttribute('href'))) { 518 links[i].target = '_blank'; 519 } 520 } 521})(); 522// adds .remark-code-has-line-highlighted class to <pre> parent elements 523// of code chunks containing highlighted lines with class .remark-code-line-highlighted 524(function(d) { 525 const hlines = d.querySelectorAll('.remark-code-line-highlighted'); 526 const preParents = []; 527 const findPreParent = function(line, p = 0) { 528 if (p > 1) return null; // traverse up no further than grandparent 529 const el = line.parentElement; 530 return el.tagName === "PRE" ? el : findPreParent(el, ++p); 531 }; 532 533 for (let line of hlines) { 534 let pre = findPreParent(line); 535 if (pre && !preParents.includes(pre)) preParents.push(pre); 536 } 537 preParents.forEach(p => p.classList.add("remark-code-has-line-highlighted")); 538})(document);</script>
538 539
540<script> 541slideshow._releaseMath = function(el) { 542 var i, text, code, codes = el.getElementsByTagName('code'); 543 for (i = 0; i < codes.length;) { 544 code = codes[i]; 545 if (code.parentNode.tagName !== 'PRE' && code.childElementCount === 0) { 546 text = code.textContent; 547 if (/^\\\((.|\s)+\\\)$/.test(text) || /^\\\[(.|\s)+\\\]$/.test(text) || 548 /^\$\$(.|\s)+\$\$$/.test(text) || 549 /^\\begin\{([^}]+)\}(.|\s)+\\end\{[^}]+\}$/.test(text)) { 550 code.outerHTML = code.innerHTML; // remove <code></code> 551 continue; 552 } 553 } 554 i++; 555 } 556}; 557slideshow._releaseMath(document); 558</script>
558 559<!-- dynamically load mathjax for compatibility with self-contained -->
560<script> 561(function () { 562 var script = document.createElement('script'); 563 script.type = 'text/javascript'; 564 script.src = 'https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-MML-AM_CHTML'; 565 if (location.protocol !== 'file:' && /^https?:/.test(script.src)) 566 script.src = script.src.replace(/^https?:/, ''); 567 document.getElementsByTagName('head')[0].appendChild(script); 568})(); 569</script>
569 570 </body> 571</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.