forked from rajag0pal/rajag0pal.github.io
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathAIBIT-details.html
More file actions
602 lines (447 loc) · 40.2 KB
/
AIBIT-details.html
File metadata and controls
602 lines (447 loc) · 40.2 KB
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
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta content="width=device-width, initial-scale=1.0" name="viewport">
<title>Rajagopal's Portfolio</title>
<meta content="" name="description">
<meta content="" name="keywords">
<!-- Favicons -->
<link href="assets/img/favicon.png" rel="icon">
<link href="assets/img/apple-touch-icon.png" rel="apple-touch-icon">
<!-- Google Fonts -->
<link href="https://fonts.googleapis.com/css?family=Open+Sans:300,300i,400,400i,600,600i,700,700i|Raleway:300,300i,400,400i,500,500i,600,600i,700,700i|Poppins:300,300i,400,400i,500,500i,600,600i,700,700i" rel="stylesheet">
<!-- Vendor CSS Files -->
<link href="assets/vendor/aos/aos.css" rel="stylesheet">
<link href="assets/vendor/bootstrap/css/bootstrap.min.css" rel="stylesheet">
<link href="assets/vendor/bootstrap-icons/bootstrap-icons.css" rel="stylesheet">
<link href="assets/vendor/boxicons/css/boxicons.min.css" rel="stylesheet">
<link href="assets/vendor/glightbox/css/glightbox.min.css" rel="stylesheet">
<link href="assets/vendor/swiper/swiper-bundle.min.css" rel="stylesheet">
<!-- Template Main CSS File -->
<link href="assets/css/style.css" rel="stylesheet">
<script src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.7/MathJax.js?config=TeX-MML-AM_CHTML"></script>
<!-- =======================================================
* Template Name: MyResume - v4.9.2
* Template URL: https://bootstrapmade.com/free-html-bootstrap-template-my-resume/
* Author: BootstrapMade.com
* License: https://bootstrapmade.com/license/
======================================================== -->
</head>
<body>
<!-- ======= Mobile nav toggle button ======= -->
<!-- <button type="button" class="mobile-nav-toggle d-xl-none"><i class="bi bi-list mobile-nav-toggle"></i></button> -->
<i class="bi bi-list mobile-nav-toggle d-lg-none"></i>
<!-- ======= Header ======= -->
<header id="header" class="d-flex flex-column justify-content-center">
</header><!-- End Header -->
<!-- ======= Services Section ======= -->
<section id="services" class="services">
<div class="container" data-aos="fade-up">
<div class="row">
<div style="text-align: left; width: 40%; margin-bottom: 0%; margin-top: 5%;">
<h2 style="color:#085a98;">AI in Biotechnology</h2>
<h2>An AI-based framework and data-driven methodology for post-PCR High-Resolution Melting Analysis (HRMA)</h2>
</div>
<div style="width:20%; ">
<picture style="margin-left: 30%; ">
<img src="assets/img/AIBITbanner.png" style="width:700px; height:300px; padding-top: 10%; padding-right: 35%;">
</picture>
</div>
</div>
<div class="icon-box iconbox-teal" data-aos="zoom-in" data-aos-delay="300" style="width: 100%; margin-top: 20px;margin-bottom: 20px;">
<h2 style="font-family: 'Poppins', sans-serif; font-size: 24px; color: #f8893f; margin-bottom: 20px; text-align: left;">Empowering Molecular Analysis and Diagnosis with AI</h2>
<p style="text-align: justify; font-size: 14px;">AI in Biotechnology paves the way for an innovative framework for automated analysis of HRM data. By harnessing cutting-edge computational techniques such as Machine Learning, Signal Processing, and Deep Learning, this project enables predictive analysis of HRM data from clinical samples. It has been meticulously designed and discussed the fundamental principles for processing, analyzing, and interpreting HRM data for a representative set of molecular targets. This pioneering approach empowers technicians and clinicians with accurate reporting and enhances diagnostic capabilities. With "AI in Biotechnology," the field of molecular analysis and diagnosis is poised for a significant transformation. This project not only bridges the gap in molecular assays for disease detection but also ensures reliable and automated analysis of HRM data. By harnessing the power of advanced computational techniques, healthcare professionals can make faster, more accurate diagnoses, ultimately saving lives.</p>
<!-- <img src="assets/img/table.png" style="width: 500px; height: 200px; margin-top: 50px;"> -->
</div>
<!-- <div class="section-title" style="margin-top: 20px;">
<p style="text-align: left;">PyHRM is a python based library for processing <strong>High Resolution Melting (HRM)</strong> data, especially, DNA melting signals to extact features like <strong>'Melting Temperatures', 'Take-off and Touch-down points of melting signal</strong> (Temperature at which peak start rising and temperature at which peak falls down)',<strong>'Peak prominences'</strong>,and <strong>'Area Under the curve'</strong>.
Additionally, the library offers interactive visualization for DNA melting singal and <strong>vision based filtering</strong>, to eliminate noisy signals from the data and provides only genuine peaks with all the above mentioned features.</p>
</div> -->
<div class="row">
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="width:100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">What is DNA Melting?</a></h4>
<br>
<div class = "row">
<p style="text-align: justify; font-size: 12px; width: 40%; margin-left: 5%;">DNA melt signals are the important output of PCR experiments, as they provide
information about the characteristics of the amplified DNA. As the temperature increases, the
double-stranded DNA begins to denature into single-strand, and the DNA-binding dye will
dissociate from the DNA, causing a decrease in fluorescence. The temperature at which half of
the double-stranded DNA is denatured is called the melting temperature (Tm).
<br>
<!-- <div data-aos="zoom-in" data-aos-delay="100" style="width: 400px; height: 250px; margin-top:5%;"> -->
<!-- </div> -->
<br>
Each pathogen will be having different DNA melting temperatures and their respective
signals will also possess different shapes and sizes. Usually, DNA melting signals are in bell-shaped curve with peaks, which denotes the melting point or the melting temperature of the
DNA. Interpretation will be made through visual inspection, performed on such signal’s shape,
peaks, and size. </p>
<video class="demo" id="videoPlayer" playsinline style="margin-top: 2%; margin-left: 15px; width:50%;" controls>
<source src="assets/img/demoPCR.mp4" type="video/mp4">
</video>
<p style="margin-left: 30%;"><a href="http://www.youtube.com/@SiemensHealthineers">Video Credit: @SiemensHealthineers</a></p>
</div>
</div>
</div>
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 3%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Signal Processing in DNA melting</a></h4>
<br>
<div class = "row">
<img src="assets/img/melt.png" style="width:40%;margin-left: 8%;">
<img src="assets/img/peak.png" style="width:40%; margin-left: 3%;">
<p style="text-align: justify; font-size: 12px; margin-left: 9%;width: 80%; margin-right: 3%; margin-top: 4%;" >To process the DNA melting signal, several techniques from signal processing can be
used. For example, noise reduction techniques, such as filtering and averaging, can be used to
remove random fluctuations in the signal and improve its accuracy. Signal processing
techniques can also be used to extract features from the melting curve, such as the melting
temperature, the shape of the curve, and the width of the transition region. These features can
then be used to compare different DNA samples or to detect mutations and structural variations
in the DNA molecule. Another important aspect of DNA melting signal processing is the use
of mathematical models to describe the melting process. Signal processing techniques play a
critical role in analysing and interpreting DNA melting signals, and can provide valuable
insights into the properties of the DNA molecule and its interactions with other molecules.
</p>
</div>
</div>
</div>
<div class="section-title" style="font-size: 20px; margin-top: 5%; margin-bottom: 0%;">
<h2>Key Features</h2>
</div>
<div class="col-lg-3 col-md-6 mt-3" data-aos="zoom-in" data-aos-delay="300" style="margin-right: 0px; margin-left: 0px;">
<div class="icon-box iconbox-orange">
<div class="icon">
<svg width="100" height="100" viewBox="0 0 600 600" xmlns="http://www.w3.org/2000/svg">
<path stroke="none" stroke-width="0" fill="#f5f5f5" d="M300,582.0697525312426C382.5290701553225,586.8405444964366,449.9789794690241,525.3245884688669,502.5850820975895,461.55621195738473C556.606425686781,396.0723002908107,615.8543463187945,314.28637112970534,586.6730223649479,234.56875336149918C558.9533121215079,158.8439757836574,454.9685369536778,164.00468322053177,381.49747125262974,130.76875717737553C312.15926192815925,99.40240125094834,248.97055460311594,18.661163978235184,179.8680185752513,50.54337015887873C110.5421016452524,82.52863877960104,119.82277516462835,180.83849132639028,109.12597500060166,256.43424936330496C100.08760227029461,320.3096726198365,92.17705696193138,384.0621239912766,124.79988738764834,439.7174275375508C164.83382741302287,508.01625554203684,220.96474134820875,577.5009287672846,300,582.0697525312426""></path>
</svg>
<i class="bx bx-vertical-top"></i>
</div>
<h4><a href="">Peak</a></h4>
<p></p>
</div>
</div>
<div class="col-lg-3 col-md-6 mt-3" data-aos="zoom-in" data-aos-delay="100" style="margin-right: 0px; margin-left: 0px;">
<div class="icon-box iconbox-teal" style="margin-left: 20px;">
<div class="icon">
<svg width="100" height="100" viewBox="0 0 600 600" xmlns="http://www.w3.org/2000/svg">
<path stroke="none" stroke-width="0" fill="#f5f5f5" d="M300,566.797414625762C385.7384707136149,576.1784315230908,478.7894351017131,552.8928747891023,531.9192734346935,484.94944893311C584.6109503024035,417.5663521118492,582.489472248146,322.67544863468447,553.9536738515405,242.03673114598146C529.1557734026468,171.96086150256528,465.24506316201064,127.66468636344209,395.9583748389544,100.7403814666027C334.2173773831606,76.7482773500951,269.4350130405921,84.62216499799875,207.1952322260088,107.2889140133804C132.92018162631612,134.33871894543012,41.79353780512637,160.00259165414826,22.644507872594943,236.69541883565114C3.319112789854554,314.0945973066697,72.72355303640163,379.243833228382,124.04198916343866,440.3218312028393C172.9286146004772,498.5055451809895,224.45579914871206,558.5317968840102,300,566.797414625762"></path>
</svg>
<i class="bx bx-horizontal-right"></i>
</div>
<h4><a href="">Prominence</a></h4>
</div>
</div>
<div class="col-lg-3 col-md-6 mt-3" data-aos="zoom-in" data-aos-delay="100" style="margin-right: 0px; margin-left: 0px;">
<div class="icon-box iconbox-red">
<div class="icon">
<svg width="100" height="100" viewBox="0 0 600 600" xmlns="http://www.w3.org/2000/svg">
<path stroke="none" stroke-width="0" fill="#f5f5f5" d="M300,532.3542879108572C369.38199826031484,532.3153073249985,429.10787420159085,491.63046689027357,474.5244479745417,439.17860296908856C522.8885846962883,383.3225815378663,569.1668002868075,314.3205725914397,550.7432151929288,242.7694973846089C532.6665558377875,172.5657663291529,456.2379748765914,142.6223662098291,390.3689995646985,112.34683881706744C326.66090330228417,83.06452184765237,258.84405631176094,53.51806209861945,193.32584062364296,78.48882559362697C121.61183558270385,105.82097193414197,62.805066853699245,167.19869350419734,48.57481801355237,242.6138429142374C34.843463184063346,315.3850353017275,76.69343916112496,383.4422959591041,125.22947124332185,439.3748458443577C170.7312796277747,491.8107796887764,230.57421082200815,532.3932930995766,300,532.3542879108572"></path>
</svg>
<i class="bx bx-vertical-bottom"></i>
</div>
<h4><a href="">Width</a></h4>
<p></p>
</div>
</div>
<div class="col-lg-3 col-md-6 mt-3" data-aos="zoom-in" data-aos-delay="100" style="margin-left: 0px;margin-right: 0px;">
<div class="icon-box iconbox-yellow">
<div class="icon">
<svg width="100" height="100" viewBox="0 0 600 600" xmlns="http://www.w3.org/2000/svg">
<path stroke="none" stroke-width="0" fill="#f5f5f5" d="M300,503.46388370962813C374.79870501325706,506.71871716319447,464.8034551963731,527.1746412648533,510.4981551193396,467.86667711651364C555.9287308511215,408.9015244558933,512.6030010748507,327.5744911775523,490.211057578863,256.5855673507754C471.097692560561,195.9906835881958,447.69079081568157,138.11976852964426,395.19560036434837,102.3242989838813C329.3053358748298,57.3949838291264,248.02791733380457,8.279543830951368,175.87071277845988,42.242879143198664C103.41431057327972,76.34704239035025,93.79494320519305,170.9812938413882,81.28167332365135,250.07896920659033C70.17666984294237,320.27484674793965,64.84698225790005,396.69656628748305,111.28512138212992,450.4950937839243C156.20124167950087,502.5303643271138,231.32542653798444,500.4755392045468,300,503.46388370962813"></path>
</svg>
<i class="bx bx-area"></i>
</div>
<h4><a href="">AUC</a></h4>
<p></p>
</div>
</div>
<!-- <div class="col-lg-3 col-md-6 mt-3" data-aos="zoom-in" data-aos-delay="100" style="margin-right: 0px; margin-left: 0px;">
<div class="icon-box iconbox-pink">
<div class="icon">
<svg width="100" height="100" viewBox="0 0 600 600" xmlns="http://www.w3.org/2000/svg">
<path stroke="none" stroke-width="0" fill="#f5f5f5" d="M300,541.5067337569781C382.14930387511276,545.0595476570109,479.8736841581634,548.3450877840088,526.4010558755058,480.5488172755941C571.5218469581645,414.80211281144784,517.5187510058486,332.0715597781072,496.52539010469104,255.14436215662573C477.37192572678356,184.95920475031193,473.57363656557914,105.61284051026155,413.0603344069578,65.22779650032875C343.27470386102294,18.654635553484475,251.2091493199835,5.337323636656869,175.0934190732945,40.62881213300186C97.87086631185822,76.43348514350839,51.98124368387456,156.15599469081315,36.44837278890362,239.84606092416172C21.716077023791087,319.22268207091537,43.775223500013084,401.1760424656574,96.891909868211,461.97329694683043C147.22146801428983,519.5804099606455,223.5754009179313,538.201503339737,300,541.5067337569781"></path>
</svg>
<i class="bx bx-bot"></i>
</div>
<h4><a href="">Start & End</a></h4>
<p></p>
</div>
</div> -->
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 6%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Components of the Project</a></h4>
<br>
<div class = "row">
<img src="assets/img/components.png" style="width:35%; margin-left: 0%;">
<p style="text-align: justify; font-size: 12px; margin-left: 0%;width: 80%; margin-right: 2%; margin-top: 0%; width: 60%;" >The project has three major components for developing an AI-based framework for
automated analysis, interpretation, and data management of High-Resolution Melting (HRM)
data. These components include the data extraction tool called <strong>"EXTRACTOR,"</strong> the feature
engineering tool called <strong>"PyHRM"</strong> and the prediction tool called <strong>"Meltcurve Interpreter"</strong>.
The EXTRACTOR tool is used to extract data from the raw HRM files (for more details refer "Extractor" section in the project), while the PyHRM tool
is used for feature engineering (for more details refer "PyHRM" section in the project), which involves extracting relevant features from the HRM data.
Finally, the Meltcurve Interpreter tool uses predictive analytics and deep learning models to
interpret the extracted features and predict the presence of the intended molecular target in a
clinical sample tested (for more details refer "Meltcurve Interpreter" section in the project).
These three components work together to develop an AI-based framework that can automate
the analysis and interpretation of HRM data, allowing for faster and more accurate diagnosis
of infectious diseases. With this framework, clinicians can make more informed decisions and
plan the course of treatment for their patients.
</p>
</div>
</div>
</div>
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 3%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Proposed Framework</a></h4>
<br>
<div class = "row">
<img src="assets/img/framework.png" style="width:100%; margin-left: 0%;">
<p style="text-align: justify; font-size: 12px; margin-left: 0%;width: 80%; margin-right: 2%; margin-top: 0%; width: 60%;" >
</p>
</div>
</div>
</div>
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 3%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Data Extraction</a></h4>
<br>
<div class = "row">
<img src="assets/img/logo_E.png" style="width:35%; height:80%;margin-left: 0%;">
<p style="text-align: justify; font-size: 12px; margin-left: 0%;width: 80%; margin-right: 2%; margin-top: 0%; width: 60%;" >EXTRACTOR is a lightweight simple GUI-based application that extracts '.rex' files from the Qiagen's Rotor-Gene Q Software to the necessary '.xls' file. It's built for the users such as laboratory technicians and clinicians who handle and run PCR experiments especially in Qiagen's Rotor-Gene Q thermal cycler machine.
After the successful experiment ran in Rotor-Gene Q cycler, it produces the raw data and the users which can be only opened and analyzed via Qiagen's Q-Rex Software. If a specific run file (raw data) has to be exported into desired formats such as text(.txt), HTML Table(.html), XML(.xml), excel(.xls) given by the Qiagen Rotor-Gene Q-Rex Software. Here we automated the user role by our EXTRACTOR Software, by which you simply put the raw data file directory and desired directory to which the excel files are stored in your system, which saves time and not to burned out from this repititive task.
<br>
<img src="assets/img/extractor.png" style="width:80%; margin-left: 12%; margin-top: 5%;">
</p>
</div>
</div>
</div>
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 3%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Feature Engineering, Pre-processing and Visualization</a></h4>
<br>
<div class = "row">
<img src="assets/img/PyHRMbanner.png" style="width:20%;height: 25%;margin-left: 7%;margin-top: 15%;">
<p style="text-align: justify; font-size: 12px; margin-left: 0%; margin-right: 0%; margin-top: 0%; width: 70%;" >PyHRM is a Python-based library specifically designed for processing High Resolution Melting (HRM) data, particularly DNA melting signals. Its primary purpose is to extract various features from the data, such as "Melting Temperatures" (the temperature at which DNA strands dissociate), the "Take-off and Touch-down points" (the temperature at which the melting signal begins to rise and fall), "Peak prominences" (the prominence of individual peaks in the signal), and the "Area Under the Curve" (the integrated area under the melting curve).
To accomplish this, PyHRM makes use of the powerful SciPy library for signal processing tasks. By leveraging SciPy's efficient algorithms for filtering, peak detection, and curve fitting, PyHRM ensures accurate and reliable feature extraction from the DNA melting signals. This ensures that the analysis results are robust and trustworthy.
In addition to signal processing, PyHRM incorporates interactive visualization capabilities using the Plotly library. With Plotly, users can visually explore and analyze the DNA melting signals in an interactive manner. They can zoom in on specific regions of interest, pan across the signal, and inspect individual data points. This interactive visualization greatly enhances the user experience and enables a more comprehensive understanding of the HRM data.
Moreover, PyHRM includes vision-based filtering to eliminate noisy signals and retain only genuine peaks with the aforementioned features. By applying this filtering technique, the library ensures that the analysis is based on reliable data, thereby improving the accuracy of the results obtained from the HRM data analysis using PyHRM.
Overall, PyHRM provides a comprehensive and user-friendly environment for processing and analyzing HRM data. It combines the signal processing capabilities of SciPy with the interactive visualization offered by Plotly, enabling researchers to effectively explore and interpret DNA melting signals, and extract valuable features for further analysis and interpretation.
</p>
</div>
</div>
</div>
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 3%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Vision Based Thresholding</a></h4>
<br>
<div class = "row">
<img src="assets/img/melt CNN.png" style="width:85%;height: 25%;margin-left: 7%;margin-top: 0%;">
</div>
<p style="text-align: justify; font-size: 12px; margin-left: 7%; margin-right: 0%; margin-top: 0%; width: 85%;" >Classifying genuine and non-genuine DNA melting signals is an important task in DNA
analysis and sequencing. Convolutional Neural Networks (CNNs) have been used to address
this problem by learning to automatically extract features from the melting signals and classify
them as genuine or non-genuine.
During training, the CNN is fed with a large dataset of labelled DNA melting signals, consisting
of both genuine and non-genuine examples. The network learns to differentiate between these
two classes by adjusting the weights of its filters through a process called backpropagation.
Once trained, the CNN can be used to classify new DNA melting signals as genuine or non-genuine. By the way using CNNs to classify genuine and non-genuine DNA melting signals
will be a promising approach.
<br>
<br>
To train a CNN model, to classify DNA melting signals of “Single Peaked”, “Double Peaked”
and “Noise”, necessary training images has to be generated and must be labelled accordingly.
ConvNets, or Convolutional Neural Networks, have shown promising results in DNA analysis,
particularly in the classification of DNA melting signals. They are able to automatically extract
relevant features from the data and learn to classify the signals based on these features. In the
context of DNA signal thresholding, ConvNets can be used to generate an image of the signal
based on provided coordinates. The image is then processed through the trained neural network,
which has learned to identify and extract specific features from the signal. The network's
feature maps, which are created during training, enable it to provide a probability distribution
indicating the likelihood that the signal belongs to a particular probability density function
93
(PDF). For example, the network might output a probability distribution indicating that the
signal is more likely to belong to the PDF of "Single Peaked" than to the PDF of "Double
Peaked" or "Noise". This information can be used to threshold the signal and separate genuine
signals from noise or other artifacts.
<br>
<img src="assets/img/imagedata.png" style="width:100%; margin-left: 0%; margin-top: 2%; height: 40%;">
<br>
The Model has a validation accuracy of 83.81% and the training accuracy of 86.67%, which
looks like, the model doesn’t overfit to the data. Since it is a multiclass classification problem,
looking on the accuracy is not sufficient. The true accuracy of the model will be assessed by
looking on metrics like precision and recall. Furtherly, on combing both the metrics, f1 score
can be taken into consideration, as it is harmonic mean of both precision and recall, will
produce a significant and reliable result if the model truly performs good.
<br>
<img src="assets/img/metrics.png" style="width:100%; margin-left: 0%; margin-top: 2%; height: 40%;">
</p>
</div>
</div>
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 3%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Meningitis</a></h4>
<br>
<div class = "row">
<img src="assets/img/menign.png" style="width:30%;height: 25%;margin-left: 5%;margin-top: 0%;">
<p style="text-align: justify; font-size: 12px; margin-left: 0%; margin-right: 0%; margin-top: 0%; width: 50%;" >Meningitis is a serious infection of the meninges, the membranes covering the brain and spinal cord. It is a devastating disease and remains a major public health challenge. The disease can be caused by many different pathogens including bacteria, fungi or viruses, but the highest global burden is seen with bacterial meningitis.
Several different bacteria can cause meningitis. <strong>Streptococcus pneumoniae</strong>, <strong>Haemophilus influenzae</strong>, <strong>Neisseria meningitidis</strong> are the most frequent ones. Meningococcal meningitis can affect anyone of any age, but mainly affects babies, preschool children and young people. The disease can occur in a range of situations from sporadic cases, small clusters to large epidemics throughout the world, with seasonal variations.
<br>
<br>
Initial diagnosis of meningococcal meningitis can be made by clinical examination followed by a lumbar puncture showing a purulent spinal fluid. The bacteria can sometimes be seen in microscopic examinations of the spinal fluid. The diagnosis is confirmed by growing the bacteria from specimens of spinal fluid or blood, or by <strong>polymerase chain reaction (PCR)</strong>.
</p>
<br>
<h4 ><a style="color: rgb(0, 87, 105);" href="">Standards for Positive Samples</a></h4>
<div class="row">
<img src="assets/img/mep1.png" style="width:50%;margin-left: 0%;margin-top: 2%;">
<img src="assets/img/mep2.png" style="width:50%;margin-left: 0%;margin-top: 5%;">
</div>
</div>
</div>
</div>
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 3%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Machine Learning</a></h4>
<br>
<div class = "row">
<p style="text-align: justify; font-size: 12px; margin-left: 10%; margin-right: 0%; margin-top: 0%; width: 30%;" >
Leveraging the developed components, data has been generated for the postive samples from the Meningitis panel and mapped with the target values. More than 100+ samples were
taken and extracted 800+ observation on each class, which contributes a total of 2500+ observations. This has been used to train a multiclass classification model using tensorflow
to classify unseen signals in future, into these recognised pathogen classes.
<br>
Negative samples will be omitted automatically, since the vision based filtering helps the analysis by mapping 0.0 to all the features, if the signals are
recognised as negative, as per the training.
</p>
<img src="assets/img/datamep.png" style="width:50%;height: 70%;margin-left: 2%;margin-top: 3%;">
<br>
<br>
<p style="text-align:left;font-size: 12px; margin-left: 30%; margin-right: 0%; margin-top: 5%;"><script type="math/tex">\mathbf{feature vector} = \begin{bmatrix} \text{Tm1} \\ \text{Tstart1} \\ \text{Tend1} \\ \text{Prom1} \\ \text{Width1} \\ \text{AUC1} \\ \text{Tm2} \\ \text{Tstart2} \\ \text{Tend2} \\ \text{Prom2} \\ \text{Width2} \\ \text{AUC2} \end{bmatrix}</script> = <script type="math/tex">\begin{bmatrix} \text{79.5504} \\ \text{78.68732} \\ \text{80.05309} \\ \text{0.983805} \\ \text{35.88093} \\ \text{4.30276} \\ \text{0.0} \\ \text{0.0} \\ \text{0.0} \\ \text{0.0} \\ \text{0.0} \\ \text{0.0} \end{bmatrix}</script></p>
<br>
<br>
<p style="text-align: justify; font-size: 12px; margin-left: 10%; margin-right: 0%; margin-top: 5%; width: 30%;" >
The Model has a test accuracy of 85% and the training accuracy of 86.67%, which looks
like, the model doesn’t overfit to the data. Since it is a multiclass classification problem,
looking on the accuracy is not sufficient.
<br>
The true accuracy of the model will be assessed by looking on metrics like
precision and recall. Furtherly, on combing both the metrics, f1 score can be taken into
consideration, as it is harmonic mean of both precision and recall, will produce a significant
and reliable result if the model truly performs good
</p>
<table style="width: 40%;height: 60%;margin-top: 10%;margin-left: 5%; border-collapse: collapse;">
<tr style="border: 1px solid black;">
<th style="border: 1px solid black; padding: 8px; text-align: center;">Class</th>
<th style="border: 1px solid black; padding: 8px; text-align: center;">Precision</th>
<th style="border: 1px solid black; padding: 8px; text-align: center;">Recall</th>
<th style="border: 1px solid black; padding: 8px; text-align: center;">F1-Score</th>
<th style="border: 1px solid black; padding: 8px; text-align: center;">Support</th>
</tr>
<tr style="border: 1px solid black;">
<td style="border: 1px solid black; padding: 8px; text-align: center;">0</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">1.000</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">1.000</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">1.000</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">19</td>
</tr>
<tr style="border: 1px solid black;">
<td style="border: 1px solid black; padding: 8px; text-align: center;">1</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">0.808</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">0.913</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">0.857</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">23</td>
</tr>
<tr style="border: 1px solid black;">
<td style="border: 1px solid black; padding: 8px; text-align: center;">2</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">0.909</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">0.800</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">0.851</td>
<td style="border: 1px solid black; padding: 8px; text-align: center;">25</td>
</tr>
</table>
<img src="assets/img/cm.png" style="width:30%;height: 70%;margin-left: 35%;margin-top: 3%;">
<img src="assets/img/lossacc.png" style="width:70%;height: 60%;margin-left: 15%;margin-top: 3%;">
</div>
</div>
</div>
<div class="col-lg-4 col-md-4 d-flex align-items-stretch" data-aos="zoom-in" data-aos-delay="100" style="margin-top: 5%; width: 100%;">
<div class="icon-box iconbox-blue">
<h4 ><a style="color: rgb(0, 87, 105);" href="">Work in Progress: Open AI Fine-tuning</a></h4>
<br>
<div class = "row">
<img src="assets/img/openai.jpg" style="width:15%;height: 13%;margin-left: 5%;margin-top: 0%;">
<p style="text-align: justify; font-size: 12px; margin-left: 0%; margin-right: 0%; margin-top: 0%; width: 50%;" >
Language Models can be put into picture for advanced interoperability. This can be acheived by creating custom data, especially
text based documnets, that has conventional text used by microbiologists to interpret the signals. By the way the final result,
that come from the trained ML model, can be further eloborated using a language model, which can give us more information on the
signals through conversational text.
</p>
<p style="text-align: justify; font-size: 12px; margin-left: 20%; margin-right: 0%; margin-top: 0%; width: 50%;" >
On the other hand BioGPT is a state-of-the-art language model designed specifically for applications in the field of biology and genetics. With its extensive knowledge of biological concepts and language understanding capabilities, BioGPT has the ability to provide text-based results for a model that classifies pathogens using DNA melting signal data.
By utilizing the power of BioGPT, researchers and scientists can obtain descriptive and informative text outputs for their pathogen classification models. These text-based results can effectively communicate the findings and insights derived from the analysis of DNA melting signal data. BioGPT assists in generating clear and concise descriptions of the classification outcomes, allowing researchers to share their research outcomes and analysis with the scientific community or other relevant stakeholders.
</p>
<img src="assets/img/bio.png" style="width:25%;height: 20%;margin-left: 0%;margin-top: 3%;">
</div>
</div>
</div>
</div>
</div>
</section>
</main><!-- End #main -->
<!-- ======= Footer ======= -->
<footer id="footer">
<div class="container">
<h4>An AI-based framework and data-driven methodology for post-PCR High-Resolution Melting Analysis (HRMA)</h4>
<p>For more information</p>
<style>
.download-button {
display: inline-block;
padding: 12px 24px;
font-size: 16px;
font-weight: bold;
text-align: center;
text-decoration: none;
background-color: #1d66a6;
color: #ffffff;
border-radius: 4px;
border: none;
transition: background-color 0.3s ease;
}
.download-button:hover {
background-color: #ffffff;
}
</style>
<a href="assets/img/RAJAGOPAL Thesis.pdf" download class="download-button">Download Thesis</a>
<div class="social-links">
<!-- <a href="https://pypi.org/project/PyHRM/" class="twitter"><i class="bx bxl-python"></i></a> -->
<!-- <a href="#" class="facebook"><i class="bx bxl-facebook"></i></a>
<a href="#" class="instagram"><i class="bx bxl-instagram"></i></a>
<a href="#" class="google-plus"><i class="bx bxl-skype"></i></a>
<a href="#" class="linkedin"><i class="bx bxl-linkedin"></i></a> -->
</div>
<div class="copyright">
© Copyright <strong><span>Rajagopal S</span></strong>. All Rights Reserved
</div>
<div class="credits">
<!-- All the links in the footer should remain intact. -->
<!-- You can delete the links only if you purchased the pro version. -->
<!-- Licensing information: [license-url] -->
<!-- Purchase the pro version with working PHP/AJAX contact form: https://bootstrapmade.com/free-html-bootstrap-template-my-resume/ -->
<!-- Designed by <a href="https://bootstrapmade.com/">BootstrapMade</a> -->
</div>
</div>
</footer><!-- End Footer -->
<div id="preloader"></div>
<a href="#" class="back-to-top d-flex align-items-center justify-content-center"><i class="bi bi-arrow-up-short"></i></a>
<!-- Vendor JS Files -->
<script src="assets/vendor/purecounter/purecounter_vanilla.js"></script>
<script src="assets/vendor/aos/aos.js"></script>
<script src="assets/vendor/bootstrap/js/bootstrap.bundle.min.js"></script>
<script src="assets/vendor/glightbox/js/glightbox.min.js"></script>
<script src="assets/vendor/isotope-layout/isotope.pkgd.min.js"></script>
<script src="assets/vendor/swiper/swiper-bundle.min.js"></script>
<script src="assets/vendor/typed.js/typed.min.js"></script>
<script src="assets/vendor/waypoints/noframework.waypoints.js"></script>
<script src="assets/vendor/php-email-form/validate.js"></script>
<!-- Template Main JS File -->
<script src="assets/js/main.js"></script>
</body>
</html>