Choose in this order: name the outcome you want and how you would measure it, check whether the compound has been tested in humans for that outcome, check its legal status and whether you can verify the vial, price twelve weeks of it, and only then look at your DNA. Genetics comes last because it is the smallest lever, and the numbers say so. In a genome scan of 27,885 people taking GLP-1 weight-loss drugs, the strongest response variant was worth about 0.76 kg of extra weight loss per copy, while a model built from non-genetic factors alone (sex, drug choice, dose, and how long people had been treated) explained most of the variation between them.
We sell DNA reports, so take this from the people with the least commercial reason to say it: your genome is the fifth thing to check, not the first. In the largest study anyone has run on who responds to a peptide drug, the strongest variant found was worth about 0.76 kg of extra weight loss per copy, and a model built from non-genetic factors alone (sex, which drug, the dose, and how long you had been on it) explained most of the variation between people 1. The order that works is this: name the outcome and how you would measure it, check whether the compound has been tested in humans for that outcome, check its legal status and whether you can verify what is in the vial, then price twelve weeks of it. Only then look at what you carry. A framework that opens with a gene chart has the order backwards, because it will happily route you to a compound that has never been tested in a single person for the thing you want. This is a way to evaluate claims, not a buying guide, and almost nothing on this page is approved for human use.
That order comes from the numbers. The GLP-1 scan covered 27,885 people taking drugs such as semaglutide, the injectable weight-loss medicines, and found one hit, in GLP1R, the gene for the receptor those drugs bind. The signal was statistically strong (P = 2.9 x 10^-10) and worth roughly three quarters of a kilogram of extra weight loss per copy of the allele, meaning per copy of that version of the gene, of the two you inherit 1. Two things about that study matter when you weigh it. The participants were 23andMe research customers, and their weight loss and side effects came from a survey they filled in rather than from clinic measurements. Every author was a current or former 23andMe employee.
Filter 1: name the outcome, and the number you would measure
Start by writing one sentence: what changes, by how much, and by when. "Better recovery" is not a target. "Achilles pain during a 5 km run drops from 6 out of 10 to 2 out of 10 by week twelve" is a target, because in week twelve you will know whether it happened. This step is unglamorous and it eliminates more bad purchases than any gene chart, because if you never defined what "something" would look like, you cannot tell whether it happened.
- Fat loss: body weight and waist circumference on the same scale and tape, same day of the week. See <a href="/peptides-for/weight-loss">peptides for weight loss</a> for what has been tested against that endpoint.
- Injury and tendon pain: a pain score during one specific loaded movement, plus how long the soreness lasts afterwards. See <a href="/peptides-for/injury-recovery">peptides for injury recovery</a> and <a href="/learn/best-peptides-for-healing">our healing compound rankings</a>. If you compete in a tested sport, read the anti-doping note in Filter 3 first.
- Focus and mental clarity: one repeatable task timed the same way, not a general feeling. See <a href="/peptides-for/focus">peptides for focus</a>, or <a href="/learn/best-peptides-for-brain-fog">what has been tested for brain fog</a>.
- Skin: one photograph, same light, same distance, same time of day. See <a href="/peptides-for/skin-aging">peptides for skin aging</a>.
- Sleep, muscle or hair: the same evidence treatment is on <a href="/peptides-for/sleep">sleep</a>, <a href="/peptides-for/muscle-growth">muscle growth</a> and <a href="/peptides-for/hair-loss">hair loss</a>.
If you have not started anything yet, <a href="/learn/peptides-for-beginners">our beginners guide</a> covers the vocabulary before you spend money on the vocabulary.
Filter 2: has it been tested in people, for that outcome?
This filter cuts the hardest, and most shortlists do not survive it intact. Three tiers matter: tested in humans in a randomised trial for the outcome you want, tested in humans for a different outcome or in a different form, and tested only in animals. Compounds in the third tier are not fake, they are unfinished. Results in rodent injury models routinely fail to reproduce when the same compound reaches human trials.
| Compound | Human evidence that exists | What the research used | Evidence tier |
|---|---|---|---|
| Semaglutide | Approved by the FDA for weight management; a genome-wide scan of 27,885 self-reporting users mapped who responds and who gets sick from it 1 | 2.4 mg weekly by injection under the skin, in 112 adults with grade 3 obesity, each treated for more than four months, enrolled March 2023 to July 2024 2 | Human randomised evidence, approved drug |
| Thymosin beta-4, full length | A phase 2 randomised trial in severe dry eye reported 35.1% less eye discomfort and 59.1% less corneal staining at day 56 versus placebo drops, in a 9-patient pilot 5 | RGN-259 0.1% eye drops, six times daily for 28 days, with follow-up to day 56 | Human trial, but a different molecule and route from the injected product |
| TB-500 | None. The human data above belongs to the full-length molecule as an eye drop, not to the shorter injected fragment sold online | No published human efficacy trial of the injected fragment | Not tested in people |
| BPC-157 | A 2025 systematic review screened 544 articles and included 36: 35 preclinical, and one clinical, a retrospective uncontrolled series of 12 patients given a single injection into the knee, of whom 7 reported relief beyond six months. The same review found no clinical safety data 4 | Rodent injury models; no established human dose | Animal stage |
| GHK-Cu | Skin and wound research on topical formulations; no published randomised trial of the injected form | Topical formulations in skin studies | Topical only for the injected claims |
| Semax | Approved and marketed in Russia; never approved by the FDA or EMA, no phase 3 Western trial, and most supporting literature published in Russian and not independently replicated. In the United States it is an unapproved new drug: a Russian approval is not evidence that it is safe or effective, and it is not lawful to market or sell it here for any medical use. It is in this table to describe what the literature contains, not as an option | Intranasal use in Russian clinical practice | Approved elsewhere, unverified here |
| Ipamorelin | No approval for human use anywhere, and no published randomised human efficacy or safety trial. Reported in rodents to raise growth hormone with less of a stress-hormone rise than older compounds in its class, which is a design intent observed in animals rather than a demonstrated human safety advantage | Animal characterisation studies | Preclinical |
Notice what the BPC-157 row does not contain. The review that found 35 of 36 studies were preclinical also reported that no clinical safety data exists for it 4. For most compounds here, what is unknown is not only whether they work, but what they do to a person over months. Missing efficacy data and missing safety data are two separate holes, and the second one rarely gets mentioned in a sales page.
Human trials, approved as a drug
One compound in common use clears this bar, and it is the one your insurer and your doctor already have a process for.
Human data exists, but not for the form sold online
The trial that gets quoted used a different molecule, a different route, or a different regulator's approval than the vial in front of you.
No randomised human efficacy trial for the injected form
Interesting animal work, plausible mechanisms, and no completed trial that would tell you what happens in a person.
Already bought something before you read this? Run the filters backwards. Write the outcome and the measurement now, ask the vendor for the third-party test on your batch, and set a date to decide. If the vial is BPC-157, <a href="/learn/bpc-157-dosage-guide">what the published studies used</a> is the starting point, and <a href="/peptides/bpc-157">the compound page</a> collects the rest of the evidence in one place.
Goal first or evidence first?
The choice that decides your shortlist
Filter 3: legal status, and what is actually in the vial
Almost none of these compounds are FDA-approved for human use, and the legal picture is more restrictive than the marketing suggests. BPC-157, TB-500, injectable GHK-Cu and semax have all appeared in Category 2 of the FDA's interim list of bulk drug substances for 503A compounding, meaning the list of ingredients a pharmacy is allowed to mix into a prescription for one named patient. Category 2 is the label the agency uses for substances that raise significant safety risks. Several compounds in this catalogue are also covered by FDA import alerts for unapproved new drugs, which lets shipments be detained without physical examination.
That list is revised periodically, nominations are withdrawn and re-reviewed, and the FDA's Pharmacy Compounding Advisory Committee has voted on further peptides without the agency adopting its recommendations. None of that movement authorises anyone to compound, sell or prescribe these for human use, and any addition to the permitted list still requires the FDA and the Secretary of Health and Human Services to act. Check the FDA's own <a href="https://www.fda.gov/drugs/human-drug-compounding/bulk-drug-substances-nominated-use-compounding">bulk drug substances page</a> rather than a vendor's summary, or this one. Most of this catalogue is unapproved, unauthorised for compounding, and sold as research chemicals labelled not for human use.
One more status check if you compete. BPC-157, TB-500 and growth hormone secretagogues such as ipamorelin are prohibited under the World Anti-Doping Code, in and out of competition. If you are tested in a sanctioned sport, this filter ends the decision regardless of what the other four say, so read the current WADA prohibited list before anything else.
Filter 4: what the experiment costs, before you start it
Price the whole experiment, not the vial: twelve weeks of supply, plus the one measurement that would tell you whether anything happened. If the measurement costs more than you are willing to spend, you are pricing a feeling rather than a result. Twelve weeks is the usual window because published protocols tend to run and then stop to reassess rather than continue indefinitely, and our <a href="/learn/peptide-cycling-protocol">cycling protocol guide</a> covers how on and off periods are normally structured. Vial prices at research-chemical vendors sit lower than most people expect, which is part of why the evidence filter matters more than the cost filter.
Affiliate disclosure, before the prices rather than after them: PeptidesDNA earns a commission on purchases made through the vendor links on our <a href="/compare">comparison page</a>. Those vendors sell material labelled for laboratory research use only, not for human consumption. Listing a price is not an endorsement, not a recommendation, and not a statement that buying or using any of these compounds is lawful. We do not sell compounds ourselves.
| Compound | Tracked vendor list price per vial, 5 August 2026 | Note |
|---|---|---|
| Ipamorelin | $17.56 | Vial sizes differ between compounds, so the price per milligram is not comparable across rows |
| TB-500 | $23.96 | The injected fragment, not the full-length molecule used in the eye-drop trial 5 |
| Selank | $25.95 | No FDA or EMA approval |
| BPC-157 | $39.99 | Animal-stage evidence; a 2025 review found 1 clinical study out of 36 included, and no clinical safety data 4 |
| Semax | $73.95 | Approved in Russia only; an unapproved new drug in the United States |
A price is not a permission slip. Everything in that table is sold labelled not for human use, and the evidence tiers earlier in this piece are unaffected by what we earn. If you do inject something anyway, stop and seek medical care for a spreading red, hot or painful injection site, fever, chest pain, trouble breathing, or swelling of the face or throat.
Filter 5: what your DNA adds, and how much
Your DNA belongs at the end of this process because it adjusts expectations inside a shortlist you have already built. It does not build the shortlist, and on the current evidence it does not set a dose for anything discussed here.
That study is the best-powered look anyone has taken at who responds to a peptide drug, and it is worth reading as a ceiling rather than a floor. It found a statistically strong hit in GLP1R, and the effect was under a kilogram per allele copy while non-genetic factors carried most of the explained variation 1. In the same study, the signals for greater weight loss and for nausea or vomiting most likely trace to one variant: pairwise co-localization was high, and multi-trait co-localization put it at a 72.6% posterior probability, so the version of the gene tied to a stronger response is probably also tied to more gut side effects. A separate nausea signal in GIPR, a related gut hormone receptor, showed up only in people taking tirzepatide 1. A variant that predicts you will do well is not a variant that predicts you will feel fine.
A smaller prospective study gives the practical version of the same idea. Among 112 adults taking semaglutide 2.4 mg weekly, those carrying two A copies at rs6923761, an rs number being nothing more than a street address for one specific letter of DNA inside the GLP1R gene, lost 1.64% of body weight per month, against 1.04% per month in people carrying at least one G copy (p = 0.03). The A/A group was 8% of the sample, nine people out of 112 2. That is a difference worth noting from a small study, and both halves of that sentence matter. Note also that this is a different variant from the one the big scan found: the scan's hit was rs10305420, which changes a single building block of the receptor, while the 112-person study tested rs6923761. Two designs, two variants, one gene, so do not read the two effect sizes as describing one allele. What it changes if you are A/A: not your dose, and not whether you are a candidate. It nudges how much loss you might expect in a good month, and, on the larger study, how rough the first few weeks may feel.
For everything else in this catalogue, no compound covered on this site has a validated pharmacogenomic dosing rule, the kind of published, tested guideline that tells a prescriber to change a dose of warfarin or clopidogrel based on a genotype. Genetics narrows a shortlist and sets expectations. It does not set a dose.
There is a second reason to keep DNA at step five, which is that consumer raw data is a starting point rather than a result. When raw files from 49 people who had used direct-to-consumer genetic tests were sent to a clinical laboratory for confirmation, 40% of the variants reported in those files could not be confirmed, and in eight further cases the variant was real but a third-party interpretation service had described its risk incorrectly 3.
23andMe's v5 chip reads roughly 640,000 preselected positions out of about 3 billion base pairs, so it can tell you what you carry at a spot it looked at and absolutely nothing about a spot it did not.
That coverage ceiling is explained in full in <a href="/learn/23andme-peptide-guide">our guide to what a 23andMe raw file contains</a>, including which chip versions carry which positions and what a no-call is.
Two worked examples show the range of what a variant is worth. In the cognitive case, a single genotype flipped the direction of a drug effect in a double-blind trial, which is about as usable as this field gets, and <a href="/learn/brain-fog-peptides-comt-bdnf-mthfr">the piece on feeling foggier on a nootropic peptide</a> works through it. In the connective-tissue case, the most-tested collagen variant tracks how likely you are to injure a ligament rather than how fast you heal one, and the athlete data behind even that has weakened, which <a href="/learn/col5a1-connective-tissue-peptides">the piece on slow tendon healing</a> lays out. Same category of evidence, opposite practical value.
There is also a well-documented case of a popular variant test that a professional body says should not be ordered at all. The American College of Medical Genetics and Genomics concluded that MTHFR polymorphism testing, MTHFR being the enzyme that recycles folate, has minimal clinical utility and should not be part of a routine workup for clotting risk or recurrent pregnancy loss, a position its board reaffirmed in April 2020 6. Popular does not mean useful.
What our report does, and what it does not
Our $99 report reads a 144-marker panel from a file you already have, calls 8 pharmacogenes, scores 15 biological pathways, and ranks 39 peptides against what you carry, with a human quality-checking it before delivery, within 24 hours. That reviewer is not your clinician, gives no medical advice, and makes no recommendation about whether you should take anything. The report describes associations, which is not the same thing as predicting your response, and it does not produce a dose. It also cannot resolve CYP2D6 or CYP3A4, two of the liver's main drug-processing genes, because reading them correctly means detecting whole duplicated or deleted stretches of DNA that a consumer chip never looks at, so any service claiming to call those from a 23andMe file is overreaching.
<a href="/methodology">Our methodology page</a> shows exactly how the scoring works, including where the evidence behind a marker is thin. If you are comparing services, our <a href="/learn/selfdecode-peptide-report-review">review of SelfDecode's peptide report</a> shows what a competing report covers and where it overreaches.
Do peptides work differently for different people?
Yes, and the largest reasons are mostly not genetic. The GLP-1 scan already ranked them: sex, drug, dose and duration together carried most of the explained variation, well ahead of the strongest genetic hit 1. Add what is physically in the vial, whether it was reconstituted and stored properly, sleep, training load, and how long you gave it before deciding, and genetics ends up as one input among many rather than the master key. It is an input worth having, and it is the last one you should reach for.
How to run all five filters in one sitting
- Write the outcome and the numberOne sentence: what changes, by how much, measured how, by when. If you cannot finish the sentence, stop here.
- Search for human trials before you search for protocolsGo to pubmed.ncbi.nlm.nih.gov and search the compound name plus "randomized" plus your outcome. If the only hits are rats, you have your answer before you read a single protocol.
- Check the regulatory status, the sport status, and the supplyConfirm whether the compound is FDA-approved for anything, whether it is on the WADA prohibited list, and what the seller will actually show you about the batch. A vendor certificate of analysis is a purity claim, not a sterility or endotoxin test, and the same investigations that found problems in research vials also found certificates issued for product that was never tested. A certificate is a reason to keep asking questions, not a clearance.
- Price twelve weeks, including the measurementSupply cost plus the blood test, photograph or timed task that will tell you whether anything happened. Budget the measurement first, because it is the part people skip.
- Bring your DNA in lastAsk of any variant: does knowing this change my expectation, my monitoring, or my shortlist? If it changes none of the three, it is a fact about you and not a decision.
- Take the shortlist to a clinician, not to a vendorNothing above is a recommendation to use any compound. Only a licensed prescriber who knows your history, your medications and your family history can judge whether anything on your list is appropriate, and for most of these compounds the clinical answer today is that there is not enough human data to say. If you inject anything, seek medical care for a spreading red, hot or painful injection site, fever, chest pain, trouble breathing, or swelling of the face or throat.
Curious what your own variants add once you have done the first four filters? Upload the raw DNA data you already have and get all 39 peptides scored against your 144-marker panel, quality-checked by a human before it reaches you. The report is educational: it describes genetic associations, is not medical advice, does not diagnose or treat any condition, does not recommend that you take any compound, and does not produce a dose.
Get your DNA report- Genetics is filter five, not filter one
Whether a compound helps you is decided mostly by what you want it for and whether it has ever been tested in a person for that. A variant adjusts your expectation inside a shortlist you have already built; it cannot build the shortlist, and on current evidence it does not set a dose for anything on this site.
- Association is not prediction
The strongest response variant anyone has found for a peptide drug was worth about 0.76 kg of extra weight loss per copy across 27,885 people, and a model built from non-genetic factors alone explained most of the variation the study could account for. A genotype shifts the odds across a group; it does not forecast what will happen to one person.
- The vial is a bigger unknown than your genome
Research-chemical vials are made outside the rules that govern medicines, and nobody is required to confirm that the label matches the contents or that the contents are sterile. Independent testing has repeatedly found mislabelled, misdosed and contaminated product. Before blaming a gene for a compound doing nothing, account for what was physically in the syringe.
Sources6
- Genetic predictors of GLP1 receptor agonist weight loss and side effects. Nature, 2026 (genome-wide association study, n = 27,885; self-reported outcomes in 23andMe research participants; all authors current or former 23andMe employees).
- A GLP1R gene variant and sex influence the response to semaglutide treatment in patients with severe obesity. Obesity (Silver Spring), 2025 (prospective, n = 112).
- Tandy-Connor S et al. False-positive results released by direct-to-consumer genetic tests highlight the importance of clinical confirmation testing for appropriate patient care. Genetics in Medicine, 2018 (n = 49).
- Vasireddi N et al. Emerging use of BPC-157 in orthopaedic sports medicine: a systematic review. HSS Journal, 2025 (544 articles screened, 36 included: 35 preclinical, 1 clinical; no clinical safety data reported).
- Sosne G et al. Thymosin beta4 significantly improves signs and symptoms of severe dry eye in a phase 2 randomized trial. Cornea, 2015 (n = 9).
- Hickey SE, Curry CJ, Toriello HV. ACMG practice guideline: lack of evidence for MTHFR polymorphism testing. Genetics in Medicine, 2013;15(2):153-156 (reaffirmed by the ACMG Board, April 2020).
This article is for informational and educational purposes only. It is not medical advice and does not diagnose, treat, cure, or prevent any disease. Consult a qualified healthcare professional before starting any peptide protocol. Individual results vary. Some outbound links are affiliate links, at no extra cost to you.
Frequently asked questions
Which peptides should I take?
Nobody can answer that from a gene chart, and the process runs in five steps: write down the outcome you want and how you would measure it in twelve weeks, check whether the compound has been tested in humans for that outcome, check its legal status and whether you can verify the vial, price the whole twelve weeks including the measurement, and only then look at your DNA. Applied to the popular list, that process leaves semaglutide with human randomised evidence, puts TB-500 and semax in a category where the human data belongs to a different molecule or a different regulator, and leaves BPC-157 at the animal stage, with 35 of 36 studies in a 2025 systematic review being preclinical and no clinical safety data at all. Everything here is education, and the decision belongs with a clinician.
How do I know which peptide is right for me?
Build the shortlist from your goal and cut it with the evidence, and when the two conflict, let the evidence win, because a compound with no human data cannot deliver a goal no matter how well you wrote the goal. Practically: search PubMed for the compound name plus "randomized" plus your outcome before you read a single protocol. If the only results are rodent studies, you have your answer. And whatever survives that cut is a shortlist to bring to a clinician, not a purchase decision, because for most compounds on this page the evidence does not yet support human use at all.
Do peptides work differently for different people?
Yes, and the biggest reasons are usually not genetic. What is physically in an unregulated vial, whether it was mixed and stored properly, the dose, how long you gave it, your sex and which drug you were on all move the result more than any variant identified so far. The largest study of the question, a genome scan of 27,885 people on GLP-1 drugs, found that a model built from non-genetic factors alone explained most of the variation it could account for, while the strongest genetic hit was worth roughly 0.76 kg of extra weight loss per copy.
Can a DNA test tell me which peptide to take?
No test on the market can tell you which peptide to take, and any that claims to is overreaching. A DNA report describes associations, which means it can suggest which compounds are worth considering first and where you might expect a smaller effect or more side effects, on evidence that ranges from strong (the GLP-1 receptor gene) to thin (most of the rest). It cannot set a dose, and consumer raw data is a starting point rather than a result: when 49 people sent their raw files for clinical confirmation, 40% of the flagged variants could not be confirmed.
This article is for informational and educational purposes only. It is not medical advice and does not diagnose, treat, cure, or prevent any disease. Consult a qualified healthcare professional before starting any peptide protocol. Individual results vary. Some outbound links are affiliate links, at no extra cost to you.