Whole genome sequencing (WGS) + artificial intelligence (AI) - A revolution?

deleder2k

Established Member (Voting Rights)
Any updates? I think WGS is extremely interesting for, especially with the use of AI. A new world. I imagine that more and more hospitals will use it as WGS is becoming very cheap, and with the AI revolution.

How many could find out why they are sick? Either through a definitive pathogenic variant or variants that causes and drives the disease, or through findings that may lay the groundwork for us to develop ME. I saw a new study which found rare/pathogenic variants linked with mitochondrial disease.

I myself did a WGS with Nebula (now DNA complete). I chose 30x which was more expensive than 1x. As I understand it the genome is only read once with 1x, and those results can be trusted. I did a muscle and metabolic gene panel with my hospital. They used 20x.

Could WGS + AI be a gamechanger? I myself found several suggested pathogenic variants, several in genes that are crucial for the mitochondria and for the production of ATP. I contacted several researchers who think I suffer from an oligopolic mitochondrial disease which also is worsened by a non dystrophic myotonia (channelopathy) which leads to buildup of lactic acid, muscle cramps, muscle pain, PEM.

One researcher found one of my variants to be so interesting that he is doing a study with genetically edited mice eith my variant (which is very rare).
I asked Perplexity about WGS+AI:

Diagnostic Applications
Small-scale WGS studies have shown remarkable diagnostic yield. One analysis of 18 ME/CFS patients identified disease-associated variants in 89% of cases, primarily in ion transport channels and mitochondrial pathways. This led to new treatment options in 83% of patients, with 57% showing substantial clinical improvement when treatments targeted identified pathways.[neuroneeds]
For mitochondrial diseases that overlap phenotypically with ME/CFS, WGS achieved 53.7% diagnostic rates and enabled precise treatment guidance, including disease-specific therapies and avoidance of contraindicated medications. The comprehensive bigenomic sequencing approach (nuclear and mitochondrial DNA) proved superior to targeted panels or exome sequencing.[pmc.ncbi.nlm.nih]
Patient Stratification
Combinatorial analysis identified 15 distinct ME/CFS patient subgroups with shared disease etiologies. For example, Subgroup 9 (27% of patients) involves mitochondrial respiration regulation via AKAP1, while Subgroup 15 involves neurotransmitter pathways through SLC6A11. This stratification enables precision medicine by matching patients to mechanism-specific therapies.[pmc.ncbi.nlm.nih +2] For individual patients, particularly those with suspected mitochondrial or specific monogenic contributions to their ME/CFS phenotype, WGS may provide actionable diagnostic information now. For the broader ME/CFS population, the technology is establishing the biological foundation necessary for future precision medicine approaches but requires several more years of development before routine clinical application.[ainvest +3] Sources:
NeuroNeeds - “Whole Genome Sequencing in 18 patients with Chronic Fatigue” (December 2022)[neuroneeds]
PMC/NIH - “Use of Whole-Genome Sequencing for Mitochondrial Disease Diagnosis” (August 2022)[pmc.ncbi.nlm.nih]
PMC/NIH - “Genetic risk factors for ME/CFS identified using combinatorial analysis” (December 2022)[pmc.ncbi.nlm.nih]
Precision Life - “How does patient stratification inform biopharma in complex chronic disease?”[precisionlife]
ME/CFS Research Review - “New study links 14 genes to ME/CFS” (September 2022)[mecfsresearchreview]
PMC/NIH - “Machine learning and multi-omics in precision medicine for ME/CFS” (January 2025)[pmc.ncbi.nlm.nih]
AInvest - “Genetic Breakthrough in ME/CFS: A New Frontier for Biotech” (August 2025)[ainvest]



While ChatGPT summarized it this way:
In ME/CFS cohorts, about 2–5 % of patients who undergo full WGS may discover a definitive or likely mitochondrial/metabolic disorder, and another 10–15 % show rare variants of uncertain but high pathogenic potential that clarify biology or point to treatable pathways (e.g., OXPHOS, fatty-acid oxidation, or channelopathies).

So while not a universal solution, WGS + AI can yield clinically meaningful insights for roughly 1 in 5 ME-diagnosed patients, especially those with early, severe, or multisystem disease


What are your thoughts, and have you considered WGS? It is big decision as it is almost impossible to not figure out certain risk variants like if one is at risk for Alzheimer’s. It is also not easy to analyze and interpret the results, but it is doable for some if one can concentrate and if one is somewhat analytical.
 
Last edited:
This is not a published paper, it’s a blog by a commercial actor that sell these tests and services.

They only had «data» on treatment response for 7/18, but there is absolutely no mention of the actual data.
While data is limited, 7/7 patients in which a treatment was attempted based on DNA results showed some degree of (anecdotal) clinical improvement, with 4/7 (57%) showing substantial improvement.
Conflicts of interest:
Dr. Boles is the Chief Medical & Scientific Officer for NeuroNeeds LLC, the start-up company that makes SpectrumNeeds®, EnergyNeeds®, QNeeds®, and CalmNeeds®. As such, he may receive financial compensation based upon by efforts and/or the success of the company. You are under no obligation to purchase this or any products, whether recommended by Dr. Boles or another health care provider. As always, it is recommended that you contact your physician regarding these products and all other changes to disease management.

I think this is a clear example of the limitations of using LLMs for research purposes. You’ll drown in garbage and hallucinations.
 
I would put my money on WGS + forestglip & co!
Ha thanks. I'll keep reading my intro to genetics book.

I do think AI will be providing insights in the future because it already is now. AI means many different things.

Precision Life, that works on ME/CFS genetics, says their tools are based on AI.

DecodeME even used a machine learning algorithm called Regenie, and Google says machine learning is considered a subset of AI.

The algorithms will probably continue to get better and better.

If we're talking about chatbots, I don't know how useful they'll be for analyzing genetic data.
 
Back
Top Bottom