-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathPoster.html
540 lines (514 loc) · 20.1 KB
/
Poster.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta charset="utf-8" />
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<meta name="generator" content="pandoc" />
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Poster.utf8</title>
<!--
Font-awesome icons ie github or twitter
-->
<link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.8.1/css/all.css" integrity="sha384-50oBUHEmvpQ+1lW4y57PTFmhCaXp0ML5d60M1M7uH2+nqUivzIebhndOJK28anvf" crossorigin="anonymous">
<link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.8.1/css/brands.css" integrity="sha384-n9+6/aSqa9lBidZMRCQHTHKJscPq6NW4pCQBiMmHdUCvPN8ZOg2zJJTkC7WIezWv" crossorigin="anonymous">
<!--
Google fonts api stuff
-->
<link href='https://fonts.googleapis.com/css?family=Special Elite' rel='stylesheet'>
<link href='https://fonts.googleapis.com/css?family=Rasa' rel='stylesheet'>
<!--
Here are the required style attributes for css to make this poster work :)
-->
<style>
@page {
size: 841mm 1189mm;
margin: 0;
padding: 0;
}
body {
margin: 0px;
padding: 0px;
width: 841mm;
height: 1189mm;
text-align: justify;
font-size: 45px;
line-height: 1.05;
}
/* RMarkdown Class Styles */
/* center align leaflet map,
from https://stackoverflow.com/questions/52112119/center-leaflet-in-a-rmarkdown-document */
.html-widget {
margin: auto;
position: sticky;
margin-top: 2cm;
margin-bottom: 2cm;
}
.leaflet.html-widget.html-widget-static-bound.leaflet-container.leaflet-touch.leaflet-fade-anim.leaflet-grab.leaflet-touch-drag.leaflet-touch-zoom {
position: sticky;
width: 100%;
}
pre.sourceCode.r {
background-color: #dddddd40;
border-radius: 4mm;
padding: 4mm;
width: 75%;
margin: auto;
margin-top: 1em;
margin-bottom: 1em;
/* align-items: center; */
}
code.sourceCode.r{
background-color: transparent;
font-size: 20pt;
border-radius: 2mm;
}
code {
font-size: 25pt;
font-family: monospace;
background-color: #00b4d824;
color: #05668d;
padding: 1.2mm;
line-height: 1;
border-radius: 2mm;
}
caption {
margin-bottom: 10px;
font-size: 20pt;
font-style: italic;
}
tbody tr:nth-child(odd) {
background-color: #00b4d820;
}
.table>thead>tr>th, .table>tbody>tr>th, .table>tfoot>tr>th, .table>thead>tr>td, .table>tbody>tr>td, .table>tfoot>tr>td{
border-spacing: 0;
font-size: 40%;
border-style: none;
padding-top: 15px;
padding-bottom: 15px;
padding-right: 1em;
padding-left: 1em;
line-height: 1em;
}
table {
margin: auto;
}
th {
padding-left: 5mm;
padding-right: 5mm;
}
.caption {
font-size: 20pt;
font-style: italic;
padding-top: 0;
}
.references {
font-size: 13.7px;
line-height: 90%;
}
/* Create three unequal columns that floats next to each other */
.column {
float: left;
padding: 0px;
}
.outer {
width: 841mm;
height: calc(1189mm * (1 - 0.090 - 0.084 - 0.01) );
-webkit-column-count: 3; /* Chrome, Safari, Opera */
-moz-column-count: 3; /* Firefox */
column-count: 3;
-webkit-column-fill: auto;
-moz-column-fill: auto;
column-fill: auto;
column-gap: 0;
padding-left: 0cm;
padding-right: 0cm;
/* -webkit-column-rule-width: 50%;
-moz-column-rule-width: 50%;
column-rule-width: 50%; */
-webkit-column-rule-style: none;
-moz-column-rule-style: none;
column-rule-style: none;
-webkit-column-rule-color: black;
-moz-column-rule-color: black;
column-rule-color: black;
background-color: #ffffff;
font-family: Rasa;
margin-top: calc(1189mm * 0.090 );
padding-top: 1em;
padding-bottom: 1em;
}
span.citation {
color: #00b4d8;
font-weight: bold;
}
a {
text-decoration: none;
color: #00b4d8;
}
#title {
font-size: 125pt;
text-align: left;
margin: 0;
line-height: 98%;
border-bottom: 0;
font-weight: normal;
background: 0;
}
#author {
color: #05668d;
margin: 0;
line-height: 85%;
font-size: 1.17em;
}
#affiliation {
padding-top: 0.1em;
color: ;
font-style: italic;
font-size: 20px;
margin: 0;
}
sup {
color: #02c39a;
}
.affiliation sup {
font-size: 20px;
}
.author {
text-align: left;
}
.author sup {
font-size: 30px;
}
.author_extra {
color: #00b4d8;
margin: 0;
line-height: 85%;
font-size: 30px;
text-align: left;
}
.outer h1, h2, h3, h4, h5, h6 {
text-align: center;
margin: 0;
font-weight: bold;
}
.section h1 {
text-align:center;
padding-bottom:5px;
background:
linear-gradient(
to left,
#ffffff 1%,
#ffffff 20%,
#05668d75 33%,
#05668d 50%,
#05668d75 66%,
#ffffff 80%,
#ffffff 99%
)
left
bottom
#ffffff
no-repeat;
background-size:100% 5px ;
margin-top: 0.5em;
margin-bottom: 0.5em;
}
.outer h2 {
text-align: center;
}
.outer p, .level2 {
color: #000000;
}
.outer ol {
padding-left: 8%;
padding-right: 8%;
text-align: left;
}
.main {
width: 841mm;
height: calc(1189mm * 0.090);
position: absolute;
background-color: #05668d;
color: #ffffff90;
font-family: Special Elite;
background-image: linear-gradient(#05668d 50%,#00b4d8);
}
.main strong {
color: #ffffff;
}
.main strong > sup {
color: #ffffff;
}
.main sup {
color: #ffffff90;
}
#main-img-left {
width: 10%;
left: 0.5in;
bottom: 0.2in;
position: absolute;
}
#main-img-center {
width: 10%;
left: calc(841mm * 0.45);
bottom: 0.5in;
position: absolute;
}
#main-img-right {
width: 10%;
right: 0.5in;
bottom: 0.2in;
position: absolute;
}
.main p {
font-size: 150px;
font-family: Special Elite;
text-align: center;
margin: 0;
position: absolute;
top: 50%;
-ms-transform: translateY(-50%);
transform: translateY(-50%);
margin-left: 1em;
}
.fab {
color: #00000030;
font-size: 25px;
}
.twitter, i {
color: #00000030;
font-size: 35px;
text-decoration: none;
}
a.email {
text-decoration: none;
color: #00000030;
font-size: 35px;
}
.envelope {
color: #00000030;
font-size: 5px;
text-decoration: none;
}
.poster_wrap {
width: 841mm;
height: 1189mm;
padding: 0cm;
}
.main_bottom {
width: 841mm;
height: calc(1189mm * 0.084);
margin-top: calc(1189mm * (1 - 0.084));
position: absolute;
background-color: #05668d;
background-image: linear-gradient(#00b4d8 10%, #05668d);
}
.section {
padding-left: 10mm;
padding-right: 10mm;
}
span > #tab:mytable {
font-weight: bold;
}
.orcid img {
width: 3%;
}
.emphasis {
background-color: #008080;
color: #ffffff;
border: solid #0b2045 3mm;
margin: 1em;
padding-left: 0;
padding-right: 0;
}
.emphasis h1 {
font-weight: bold;
background: none;
background-color: #0b2045;
padding-bottom: 5mm;
padding-top: 1mm;
margin-top: -1mm;
margin-right: -1mm;
margin-left: -1mm;
}
.emphasis blockquote {
border: 0;
}
.emphasis ol {
padding: 0;
padding-left: 8%;
font-size: 100%;
font-weight: bold;
}
.emphasis p {
color: #ffffff;
}
</style>
</head>
<body>
<div class="poster_wrap">
<div class="column outer">
<div class="section">
<h3 id="author" class="author">
Gihawi, A.<sup> 1, <a class="orcid" href="https://orcid.org/0000-0002-3676-5561"><img src="https://raw.githubusercontent.com/brentthorne/posterdown/master/images/orcid.jpg"></a></sup><br>
<a class="twitter" href="https://mobile.twitter.com/AbrahamGihawi"><i class="fab fa-twitter"></i> @AbrahamGihawi</a><br>
<a class='envelope'><i class="fas fa-envelope"></i></a> <a href="mailto:[email protected]" class="email">[email protected]</a> <br>
</h3>
<h5 id="author_extra", class="author_extra">
Hurst, R.<sup>1</sup>
Leggett, R.M.<sup>2</sup>
Cooper, C.S.<sup>1</sup>
Brewer, D.S.<sup>1,2</sup>
Genomics England Research Consortium<sup>3</sup>
</h5>
<p id="affiliation" class="affiliation">
<sup>1</sup> Bob Champion Research and Education Building, University of East Anglia, Norwich, UK<br> <sup>2</sup> Earlham Institute, Norwich, UK<br> <sup>3</sup> Genomics England, London, UK
</p>
</div>
<div id="background" class="section level1">
<h1>Background</h1>
<p>The role of <em>Helicobacter pylori</em><span class="citation">[<a href="#ref-RN102">1</a>]</span> and Human papillomavirus<span class="citation">[<a href="#ref-RN138">2</a>]</span> in gastric and cervical cancer are testament to the prominent role that pathogens can play in cancer. When submitting tumours to whole genome sequencing, it is possible to indicentally sequence microbes in close proximity<span class="citation">[<a href="#ref-RN455">3</a>]</span>. We have been using the 100,000 Genomes Project as a rich resource to search for evidence of microbial DNA.</p>
<p>We benchmarked software to devise the best approach for cancer whole genome sequence metagenomics. The top performing approaches are provided in a tool called <a href="https://github.com/UEA-Cancer-Genetics-Lab/sepath_tool_UEA">SEPATH</a> <span class="citation">[<a href="#ref-RN454">4</a>]</span> which performs the following:</p>
<ul>
<li>Extracts unmapped reads from BAM files</li>
<li>Quality trimming & human read depletion</li>
<li>Metagenomic classification with Kraken<span class="citation">[<a href="#ref-RN72">5</a>]</span></li>
</ul>
<p>Additionally, we have also been investigating the taxonomy and functional potential of contigs produced by metagenomic assembly.</p>
</div>
<div id="methods" class="section level1">
<h1>Methods</h1>
<p>Non-human reads were extracted and classified using <a href="https://github.com/UEA-Cancer-Genetics-Lab/sepath_tool_UEA">SEPATH</a>. Classifications from PCR-free, fresh-frozen samples (<em>N</em>=7,775) with <span class="math inline">\(<20\)</span> reads were filtered. Taxa were removed according to published ‘<em>black lists</em>’ of common contaminants<span class="citation">[<a href="#ref-Eisenhofer2019">6</a>]</span>. Ordination was carried out with <a href="https://cran.r-project.org/web/packages/Rtsne/Rtsne.pdf">Rtsne</a> (perplexity=90, max_iter=2,000) on a matrix of Spearman’s distances created with the <a href="https://rdrr.io/rforge/ClassDiscovery/">ClassDiscovery</a> package.</p>
<p>Metagenomic assembly was carried out on non-human reads pooled by cancer type with <a href="https://github.com/voutcn/megahit">MEGAHIT</a><span class="citation">[<a href="#ref-RN269">7</a>]</span>. Taxonomic classifications of contigs were obtained with <a href="https://github.com/bbuchfink/diamond">DIAMOND</a><span class="citation">[<a href="#ref-RN204">8</a>]</span> with <a href="https://www.ncbi.nlm.nih.gov/refseq/about/nonredundantproteins/">NCBI non-redundant proteins</a>.</p>
<p>Functional potential of putative proteins was estimated using <a href="https://github.com/tseemann/prokka">Prokka</a><span class="citation">[<a href="#ref-Seemann2014">9</a>]</span> and <a href="https://github.com/ebi-pf-team/interproscan">InterProScan</a><span class="citation">[<a href="#ref-Jones2014">10</a>]</span>.</p>
</div>
<div id="sepath-results" class="section level1">
<h1>SEPATH Results</h1>
<p>Colorectal and oral cancers demonstrate the greatest median number of microbial reads. A background number of classified reads exists throughout all cancer types.</p>
<div class="figure" style="text-align: center"><span id="fig:cancertype"></span>
<img src="images/cancertypesummary.png" alt="Microbial reads in each tumour type" width="100%" />
<p class="caption">
Figure 1: Microbial reads in each tumour type
</p>
</div>
<div id="colorectal-and-oral-cancers-show-distinctive-microbial-communities" class="section level4">
<h4>Colorectal and Oral Cancers Show Distinctive Microbial Communities</h4>
<div class="figure" style="text-align: center"><span id="fig:cancertypetsne"></span>
<img src="images/krakentsne.png" alt="t-SNE plot of cancer samples using Spearman's distance coloured by tumour type. Colorectal and oral cancer are shown in green and blue respectively and separate out into clusters in the bottom half of the plot. This plot was produced using a reduced set of 652/1534 genera." width="108%" />
<p class="caption">
Figure 2: t-SNE plot of cancer samples using Spearman’s distance coloured by tumour type. Colorectal and oral cancer are shown in green and blue respectively and separate out into clusters in the bottom half of the plot. This plot was produced using a reduced set of 652/1534 genera.
</p>
</div>
<p><br></p>
</div>
</div>
<div id="assembly-results" class="section level1">
<h1>Assembly Results</h1>
<p>Assembling microbial reads within each tumour type has resulted in a total of 17.8 million contigs. The number of contigs produced by each cancer type was positively correlated with the number of reads submitted to each assembly (Spearman’s <span class="math inline">\(\rho = 0.87\)</span>)</p>
<div class="figure" style="text-align: center"><span id="fig:asscontaminants"></span>
<img src="Poster_files/figure-html/asscontaminants-1.png" alt="The number of contigs in each assembly after removing mammalian and common contaminant genera. Colorectal samples were excluded due to being prohibitively large to assemble as one pool" width="100%" />
<p class="caption">
Figure 3: The number of contigs in each assembly after removing mammalian and common contaminant genera. Colorectal samples were excluded due to being prohibitively large to assemble as one pool
</p>
</div>
<p><br></p>
</div>
<div id="functional-results" class="section level1">
<h1>Functional Results</h1>
<p>5,264 different pathways were reported across all ontologies, representing ~10% of all known metabolic pathways. This data has been made available via: <a href="https://UEA-Cancer-Genetics-Lab.github.io/Pancancer_Microbial_Pathways/">https://UEA-Cancer-Genetics-Lab.github.io/Pancancer_Microbial_Pathways/</a>.</p>
<p>This has suggested some tantalising pathways for future research such as <em>“PD-L1 expression and PD-1 checkpoint pathway in cancer”</em></p>
<p>It is hoped that this resource can provide researchers with an additional strand of evidence for a non-human pathway existing in cancer.</p>
<p>The number of pathway hits was correlated with the number of assembled contigs (Spearman’s <span class="math inline">\(\rho = 0.92\)</span>) and is therefore sensitive to the sample size. For this reason it is not advisable to investigate differences between cancer types.</p>
<div class="figure" style="text-align: center"><span id="fig:allpathways"></span>
<img src="Poster_files/figure-html/allpathways-1.png" alt="The distribution of pathway hits within all cancer types across all ontologies. The number of pathways for each ontology is demonstrated on the x-axis" width="100%" />
<p class="caption">
Figure 4: The distribution of pathway hits within all cancer types across all ontologies. The number of pathways for each ontology is demonstrated on the x-axis
</p>
</div>
</div>
<div id="conclusions" class="section level1">
<h1>Conclusions</h1>
<ul>
<li><p>SEPATH suggests limited pancancer microbial structure.</p></li>
<li><p>This may be caused by inbalances in reference genomes available.</p></li>
<li><p>Metagenomic assembly is a reference-independent approach that may reveal more about pancancer microbial structure.</p></li>
<li><p>All metabolic pathways reported have been made available for researchers.</p></li>
<li><p>This resource should be used for hypothesis generation or as preliminary evidence for a pathway in cancer.</p></li>
</ul>
</div>
<div id="ongoing-tasks" class="section level1">
<h1>Ongoing Tasks</h1>
<p>The technical difficulties of assembling colorectal data has been circumvented by dividing the pooled reads into six sets. In order to fairly compare cancer types, a single database must be created. To do this, we have concatenated all assembled contigs which contains >18 million contigs. To reduce this database and remove redundancy we have selected representative sequences by clustering with <a href="https://github.com/weizhongli/cdhit/wiki">CD-HIT</a><span class="citation">[<a href="#ref-Li2006">11</a>]</span> which has removed 8 million contigs. Each sample is currently being pseudo-aligned to this pancancer database with <a href="http://pachterlab.github.io/kallisto/">Kallisto</a><span class="citation">[<a href="#ref-Bray2016">12</a>]</span>.</p>
</div>
<div id="acknowledgements" class="section level1">
<h1>Acknowledgements</h1>
<p>This poster was created in <a href="https://github.com/brentthorne/posterdown">posterdown</a> and the code to do so is available on <a href="https://github.com/UEA-Cancer-Genetics-Lab/EACR_Bioinformatics_2021">GitHub</a></p>
<p>Thanks to <a href="http://www.genomicsengland.co.uk">Genomics England</a> including participants and staff as well as <a href="http://www.big-c.co.uk/">Big C</a> and <a href="https://prostatecanceruk.org/">Prostate Cancer UK</a> for supporting this project.</p>
<p><img src="resources/logos.png" width="100%" style="display: block; margin: auto;" /></p>
<div id="references" class="section level3 unnumbered">
<h3>References</h3>
<div id="refs" class="references">
<div id="ref-RN102">
<p>1. Plummer M, Martel C de, Vignat J, Ferlay J, Bray F, Franceschi S. Global burden of cancers attributable to infections in 2012: A synthetic analysis. The Lancet Global Health. 2016;4:e609–16.</p>
</div>
<div id="ref-RN138">
<p>2. Hibbitts S, Tristram A, Beer H, McRea J, Rose B, Hauke A, et al. UK population based study to predict impact of hpv vaccination. J Clin Virol. 2014;59:109–14. doi:<a href="https://doi.org/10.1016/j.jcv.2013.12.002">10.1016/j.jcv.2013.12.002</a>.</p>
</div>
<div id="ref-RN455">
<p>3. Magiorkinis G, Matthews PC, Wallace SE, Jeffery K, Dunbar K, Tedder R, et al. Potential for diagnosis of infectious disease from the 100,000 genomes project metagenomic dataset: Recommendations for reporting results. Wellcome Open Research. 2019;4.</p>
</div>
<div id="ref-RN454">
<p>4. Gihawi A, Rallapalli G, Hurst R, Cooper CS, Leggett RM, Brewer DS. SEPATH: Benchmarking the search for pathogens in human tissue whole genome sequence data leads to template pipelines. Genome Biol. 2019;20:208. doi:<a href="https://doi.org/10.1186/s13059-019-1819-8">10.1186/s13059-019-1819-8</a>.</p>
</div>
<div id="ref-RN72">
<p>5. Wood D, Salzberg S. Kraken - ultrafast metagenomic sequence classification using exact alignments. Genome Biol. 2014;15.</p>
</div>
<div id="ref-Eisenhofer2019">
<p>6. Eisenhofer R, Minich JJ, Marotz C, Cooper A, Knight R, Weyrich LS. Contamination in low microbial biomass microbiome studies: Issues and recommendations. Trends in Microbiology. 2019;27:105–17. doi:<a href="https://doi.org/10.1016/j.tim.2018.11.003">10.1016/j.tim.2018.11.003</a>.</p>
</div>
<div id="ref-RN269">
<p>7. Li D, Liu CM, Luo R, Sadakane K, Lam TW. MEGAHIT: An ultra-fast single-node solution for large and complex metagenomics assembly via succinct de bruijn graph. Bioinformatics. 2015;31:1674–6. doi:<a href="https://doi.org/10.1093/bioinformatics/btv033">10.1093/bioinformatics/btv033</a>.</p>
</div>
<div id="ref-RN204">
<p>8. Buchfink B, Xie C, Huson DH. Fast and sensitive protein alignment using diamond. Nature Methods. 2015;12:59–60.</p>
</div>
<div id="ref-Seemann2014">
<p>9. Seemann T. Prokka: Rapid prokaryotic genome annotation. Bioinformatics. 2014;30:2068–9. doi:<a href="https://doi.org/10.1093/bioinformatics/btu153">10.1093/bioinformatics/btu153</a>.</p>
</div>
<div id="ref-Jones2014">
<p>10. Jones P, Binns D, Chang H-Y, Fraser M, Li W, McAnulla C, et al. InterProScan 5: Genome-scale protein function classification. Bioinformatics. 2014;30:1236–40. doi:<a href="https://doi.org/10.1093/bioinformatics/btu031">10.1093/bioinformatics/btu031</a>.</p>
</div>
<div id="ref-Li2006">
<p>11. Li W, Godzik A. Cd-hit: A fast program for clustering and comparing large sets of protein or nucleotide sequences. Bioinformatics. 2006;22:1658–9. doi:<a href="https://doi.org/10.1093/bioinformatics/btl158">10.1093/bioinformatics/btl158</a>.</p>
</div>
<div id="ref-Bray2016">
<p>12. Bray NL, Pimentel H, Melsted P, Pachter L. Near-optimal probabilistic RNA-seq quantification. Nature Biotechnology. 2016;34:525–7. doi:<a href="https://doi.org/10.1038/nbt.3519">10.1038/nbt.3519</a>.</p>
</div>
</div>
</div>
</div>
</div>
<div class="main">
<p><strong>Microbial DNA in Cancer Sequence Data</strong></p>
</div>
<div class="main_bottom">
<img id="main-img-left" src=>
<img id="main-img-center" src=>
<img id="main-img-right" src=>
</div>
</div>
<!-- dynamically load mathjax for compatibility with self-contained -->
<script>
(function () {
var script = document.createElement("script");
script.type = "text/javascript";
var src = "true";
if (src === "" || src === "true") src = "https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-MML-AM_CHTML";
if (location.protocol !== "file:" && /^https?:/.test(src))
src = src.replace(/^https?:/, '');
script.src = src;
document.getElementsByTagName("head")[0].appendChild(script);
})();
</script>
</body>
</html>