Optimize Nutritional Supplements for Personalized Anti-Aging Strategy with Bioinformatics & Machine Learning
Abstract
In this blog, I will demonstrate a tailor-made strategy for optimizing a dietary supplement diet based on personal genomic data. This blog is not intended as a recommended way to treat your health problems, but rather as an example of how biohackers can use an open source bioinformatics tool for self-study.
Supplements optimization pipeline
As a first step, I passed a genetic test so that my genome was sequenced by one of the genomic test providers (in my case, Ancestry). They allow clients to upload their genome file (see example below)
I’ve used Ancestry.com, but there are several others that allow customers to upload their DNA data. The data can usually be downloaded in text format that distinguishes between personal and reference genomes.
The human genome is about 3 GB, but since we all have 99% identical DNA, your information will be provided as a file of your personal differences compared to the “reference genome.” The expected file size will be ~ 18 MB.
For those looking to explore your personal DNA in more detail, I would recommend a good blog, Building Your Personal Genomic Data Repository, which walks you through the steps to build a pipeline to explore the unique characteristics of your genomic data.
After receiving the DNA data, I compiled a list of diseases that people acquire with age. Details of these diseases can be found on the National Institutes of Health website using this URL with a unique disease identifier: https://meshb.nlm.nih.gov/record/ui?ui="MeSH unique disease identifier in the browser “
List of unique IDs for some age related disease
1. Age-Related Macular Degeneration (AMD) — D008268
2. Cataracts — D002386
3. Glaucoma — D005901
4. Chronic Obstructive Pulmonary Disease — D029424
5. Atherosclerosis — D003324
6. Cardiovascular disease — D002318
7. Stroke — D020521
8. Heart Diseases — D006331
9. Hypertension — D006973
10. Hypertension, Pulmonary — D006976
11. Myocardial Ischemia — D017202
12. Brain Ischemia — D002545
13. Cancer (Neoplasms) — D009369
14. Colorectal Neoplasms — D003123
15. Arthritis — D001172
16. Spondylarthritis — D025241
17. Osteoarthritis — D010003
18. Osteoporosis — D010024
19. Diabetes Mellitus — D003920
20. Diabetes Mellitus, Type 2 — D003924
21. Alzheimer’s disease — D000544
22. Dementia — D003704
23. Frontotemporal disorders (Frontotemporal Dementia) — D057180
24. Lewy body dementia — D020961
25. Huntington’s disease — D006816
26. Parkinson’s Disease — D010300
27. Frontotemporal Lobar Degeneration — D057174
28. Benign Prostatic Hyperplasia — D011470
29. Overactive Bladder — D053201
30. Gallbladder Diseases — D005705
Mutation Look-up table
The next step is to put together a checklist of your personal mutations that may lead to a higher risk of developing certain diseases over time.
I’ve used the Snpedia website to get information on the genomics of these diseases. Select the disease you want to study, for example osteoarthritis, and go to — https://www.snpedia.com/index.php/Osteoarthritis to find out which mutation may increase your risk of developing a particular disease at the moment. From the general link, you can navigate to a specific mutation (example: https://www.snpedia.com/index.php/Rs7639618)
Another example of a mutation that might be interesting is macular degeneration — age-related macular degeneration: https://www.snpedia.com/index.php/Rs10490924
So, you can go through the list of diseases that people can acquire in old age and study what mutations can be found in your genome, so that you have a list of potential complications that may arise over time.
As a final step, we can use the public DrugBank website to research potential substances that are recommended by supplement suppliers for alleviating the symptoms of certain diseases.(Example of “Nutraceutical” for Osteoarthritis can be found here)
Pipeline Automation
The last step is manual and requires automation. I’ve installed and trained the DRKG model to automatically generate disease-specific drug / supplement candidates. The DRKG was originally created to offer candidate drugs for COVID19, but it can be extended to general types of diseases. To do this, you will have to slightly edit the code.
1. Create new variable and initiate with a disease MeSH id
You will need to modify the “COV_disease_list” variable in the provided Jupiter notebook to identify the MeSH IDs for the diseases for which you might want to look for nutritional supplements.
Example:
Aging_disease_list = [
‘Disease::MESH:D010003’
]
Aging_disease_list = [
‘Disease::MESH:D000544’
]
2. Have your supplements list
You can create a file with a list of nutritional supplements, which you might want to examine and save as a separate file:
DRKG/drug_repurpose/Disease_clinical_trial_drugs.tsv
instead of
DRKG/drug_repurpose/COVID19_clinical_trial_drugs.tsv
Here is an example of some supplements IDs taken from DrugBunk:
1 Carnosine DB11695
2 Curcumin DB11672
3 Nadide DB14128
4 Resveratrol DB02709
5 Astaxanthin DB06543
6 Nicotinamide DB02701
7 AloeVeraLeaf DB13906
8 AlphaLipoicAcid DB00166
9 Andrographolide DB05767
10 Apigenin DB07352
11 AstragalusSynonymTragacanth DB10667
12 EGCG DB12116
13 Spirulina DB14744
14 EllagicAcid DB08846
15 Melatonin DB01065
16 Metformin DB00331
17 Naringenin DB03467
18 NicotinamideRiboside DB14933
19 PyridoxamineB6 DB11673
20 Quercetin DB04216
21 Rosemary DB14272
22 RosemaryOil DB14689
23 Sulforaphane DB12422
24 NMN DB03227
25 Allantoin DB11100
26 Fluoxetine DB00472
27 NicotinamideRiboside DB14933
28 Remdesivir DB14761
29 Hydroxychloroquine DB01611
30 Eculizumab DB01257
31 Methylprednisolone DB00959
32 Bevacizumab DB00112
33 beta-Sitosterol DB14038
34 CoQ10-Ubidecarenone DB09270
35 Pyridoxamine DB11673
36 Fluvoxamine DB00176
3. Run modified notebook for each disease you might be interested and collect list supplements.
Example: list of suggested drug/food supplements for a person from risk group to develop Osteoarthritis:
[6] Curcumin -0.15159688889980316
[9] EGCG -0.15539972484111786
[10] Andrographolide -0.15957969427108765
[24] Melatonin -0.18418636918067932
[29] Methylprednisolone -0.1930312216281891
[63] Quercetin -0.24154861271381378
[79] Astaxanthin -0.2605234980583191
[92] Nadide -0.27247798442840576
How to test your strategy
As I mentioned in the introduction to this blog. This solution is for personal exploration, not as a tool to treat your health problem. If you really want to try something, check with your doctor.
To track if supplements change anything in my condition, I used InsideTracker to periodically check all vital blood biomarkers.
Plus, microbiome testing done by companies like Viome allowed me to track if my gut microbiome changed while taking supplements.
Continuous glucose monitoring devices can be another way to monitor your glucose levels while you are taking nutritional supplements.