Evidence audit of the Atlas’s mTOR corpus
This page measures the Atlas against itself. Not how many studies it holds — that number says nothing about the field — but what kind of evidence this Atlas’s mTOR pathway map is built on: how much of it reaches a human, which parts of the map have no human study cited behind them, how many links in the map rest on a single paper, and how much of what is claimed carries a recorded boundary.
Every number below is computed from the corpus at build time, never written by hand. The method is at the bottom, so anyone can recompute them or apply the same audit to a different pathway.
What the corpus is made of
Each study carries a code for the kind of claim it can support, not for how good it is. A mechanistic paper is not worse than a trial; it answers a different question. The distribution matters because a pathway described almost entirely by molecular work supports conclusions about molecules, not about people.
| Evidence code | Studies | Share | |
|---|---|---|---|
| S — synthesis of human data | 3 | 0.6 % | |
| H — human study | 50 | 9.5 % | |
| A — animal model | 155 | 29.5 % | |
| M — molecular (cells, biochemistry, structure) | 268 | 51.0 % | |
| R — review, secondary literature | 43 | 8.2 % | |
| Off the ladder (preprint, registered trial) | 7 | 1.3 % |
53 of 526 studies — 10.1 % — are human evidence. Almost everything else is animal, cellular or theoretical (the 7 off-ladder records are preprints and registered trials, which can include human studies not yet reported). That is the single most important number on this page, and it is a fact about this collection, not a measurement of the field. Read it as a count of study LABELS: each record carries one code for the kind of study it mainly is, so a paper that mixes a human component with animal or cell work is counted once, under its main code. ORE2006 and CAR2008 are the clearest examples -- both include patient material alongside experimental work, and a single code hides that. The number is therefore a floor on human experimental components, not an exhaustive count of them.
M and R are listed apart on purpose. Both sit on the same rung of the internal grade, but a primary molecular experiment and a narrative review are not the same kind of evidence: one reports a measurement, the other reports what other papers measured. The corpus holds 268 molecular studies and 43 reviews. Where this page grades a pathway link rather than a study, the two cannot be separated, because a link can cite both and is graded by the strongest study behind it — so a link shown at that rung may rest on primary work, on a review, or on both.
Is this corpus’s human share rising over time?
If the pathway were maturing towards clinical use, the share of human work would rise over time. Here is what the corpus shows, by five-year window.
| Period | Studies | Human | Share | |
|---|---|---|---|---|
| 1975–1979 | 1 | 0 | 0 % | |
| 1990–1994 | 6 | 0 | 0 % | |
| 1995–1999 | 11 | 0 | 0 % | |
| 2000–2004 | 48 | 2 | 4 % | |
| 2005–2009 | 76 | 5 | 7 % | |
| 2010–2014 | 123 | 13 | 11 % | |
| 2015–2019 | 78 | 6 | 8 % | |
| 2020–2024 | 33 | 3 | 9 % | |
| 2025–2029 | 149 | 24 | 16 % |
Which parts of the pathway reach a human
The map holds 138 causal links, each of which must cite at least one study in the corpus. Grading every link by the strongest evidence behind it shows where the pathway is understood in people and where it is understood only in cells.
| Strongest evidence behind the link | Links | Share | |
|---|---|---|---|
| H — human study | 15 | 11 % | |
| A — animal model | 40 | 29 % | |
| M — molecular / in vitro / review | 83 | 60 % |
15 of 138 links (11 %) have any human evidence behind them; 83 (60 %) rest on mechanistic or review papers alone. This is the quantified form of what the Atlas elsewhere calls the human endpoint desert.
How much rests on one paper
A link cited by one study is not wrong. It is supported by one indexed publication in this corpus, which is a different thing, and it is the first place to look when a result fails to reproduce. Note what that count is not: it counts publications indexed here, so it does not establish that the field holds no replication, and a single paper often carries several independent experiments.
54 of 138 links (39 %) are carried by a single paper in this corpus. Some of those papers are definitive structures; others are one result nobody has repeated. The Atlas does not currently distinguish the two, and that is itself a gap.
4EBP1-LONGEVITY, 4EBP1-MITO, BISTERIC-MTORC1, CGASSTING-SENESCENCE, CHOL-LYCHOS, CR-LONGEVITY, DEPTOR-MTOR, ERK-NUCL, ERK-TSC, EVE-IMMUNE, EVE-PNET, EVE-RCC, FOXO-LONGEVITY, GLN-MTORC1-ARF1, GLN-RAG, HYPOXIA-REDD1, ISR-SALR, LARS-RAG, LEU-LARS, LYCHOS-GATOR1, MET-PKM2, METFORMIN-MTORC1, MITODYS-MTORC1, MTORC1-MAPK, MTORC1-PGC1A, MTORC1-PROSTATE, MTORC1-ROS, MTORC1-SAM, MTORC2-MAM, MTORC2-PROSTATE, MTORC2-SGK1, PDCD4-MTORC2, PDCD4-TRANSL, PKM2-GATOR2, PTEN-PI3K, RAG-TSC, RAGULATOR-CGASSTING, RAPA-LAM, ROS-MTORC1, S6K1-LONGEVITY, S6K1-PDCD4, SALR-MTORC1, SAM-SAMTOR, SAMTOR-GATOR1, SESN2-AGING, TAX1BP1-TBK1, TAX1BP1-ULK1, TBC1D7-TSC, TBK1-FNIP1, TEM-MTORC1, TEM-RCC, ULK1-AMPK, ULK1-FNIP1, VATPASE-TAX1BP1
How much carries its boundary
A claim without its conditions is a claim that has quietly been generalised. The Atlas records, for each link, the species, tissue, dose, nutrient state or duration the finding was established under — the point where it stops being universal.
138 of 138 links (100 %) carry a recorded boundary condition. The remainder are not thereby unconditional: they are links where the boundary has not yet been written down. Treating a blank as "holds everywhere" is the exact error this layer exists to prevent.
What kind of boundary
The sentence is for a reader; the tag is what makes it answerable. "Female mice only in SEL2009" is clear to anyone and invisible to a query, which is why recorded boundaries are also given a machine-readable kind — but that tagging is still partial, and the numbers below say how partial. 67 of 138 links (49 %) carry at least one kind, and 28 carry a hand-written one; the rest state their boundary in prose only. A tag is only added where the boundary sentence actually carries that condition, so the gap closes as the prose is written, not by labelling links more confidently than they were measured. The first group below describes the biology; the second records a curator’s judgement that the link is disputed; the third is computed from the state of the corpus rather than written by hand.
Boundaries on the biology
| Kind | Links | Which |
|---|---|---|
| cell-line-only the studies cited for this link used cell culture only | 10 | 4EBP1-MITO, LARS-RAG, LKB1-AMPK, MET-PKM2, MTORC1-GRB10, PKM2-GATOR2, TAX1BP1-TBK1, TAX1BP1-ULK1 and 2 more |
| diet-dependent depends on the diet the animal was on | 2 | 4EBP1-LONGEVITY, 4EBP1-MITO |
| dose-dependent depends on the dose, not merely on the drug being present | 2 | EVE-IMMUNE, RAPA-LONGEVITY |
| nutrient-state-dependent depends on whether the cell is fed or starved | 3 | 4EBP1-MITO, AKT-TSC, LYSO-MTORC1 |
| sex-specific shown in one sex only | 2 | RAPA-LONGEVITY, S6K1-LONGEVITY |
| single-species shown in one species and not tested elsewhere | 11 | 4EBP1-LONGEVITY, GHIGF1-LONGEVITY, ISR-SALR, MTORC1-PROSTATE, MTORC2-INSULINRES, MTORC2-LIPID, MTORC2-PROSTATE, RAPA-LONGEVITY and 3 more |
| tissue-specific holds in particular tissues or cell types | 8 | AKT-TSC, EVE-LAM, GLN-MTORC1-ARF1, GLN-RAG, MTORC1-PROSTATE, MTORC2-LIPID, MTORC2-PROSTATE, RAPA-MTORC2 |
Set by the curator
| Kind | Links | Which |
|---|---|---|
| contested the Atlas records the direction, mechanism or scope of this link as disputed | 8 | AMPK-ULK1, EVE-IMMUNE, GLN-MTORC1-ARF1, GLN-RAG, LARS-RAG, LEU-LARS, METFORMIN-AMPK, METFORMIN-MTORC1 |
Computed from the corpus
| Kind | Links | Which |
|---|---|---|
| reversible-on-withdrawal the effect goes away when the intervention stops | 2 | EVE-TSC, RAPA-LAM |
| single-study the whole link rests on one paper in this corpus | 54 | 4EBP1-LONGEVITY, 4EBP1-MITO, BISTERIC-MTORC1, CGASSTING-SENESCENCE, CHOL-LYCHOS, CR-LONGEVITY, DEPTOR-MTOR, ERK-NUCL and 46 more |
| time-dependent depends on how long the exposure lasted | 3 | EVE-IMMUNE, MTORC2-INSULINRES, RAPA-MTORC2 |
This is the query no other resource can answer. Reactome, KEGG, SIGNOR and Open Targets record the interaction; none of them record that it was only ever shown in one sex, or only in cell culture, in a form you can query across a whole pathway. The two time-related kinds have a page of their own: how long the exposure lasted, and what happens when it stops.
What has been tested and failed
Findings that did not replicate, and compounds that did not extend lifespan, are kept with the same visibility as positive results.
The corpus holds 8 studies whose curated category is Negative_result (HOR2021, JIB2026, MAN2021, MIL2011, POU2013, SIL2018, STE2026, STR2012). The label marks studies whose headline result was null; a trial with a null pre-registered primary endpoint may carry another category (MOE2025, the PEARL trial, is filed as Human). The pathway map also has a dedicated sign for a tested-and-null link (no-effect), and it is currently used 0 times. That gap is real: null relationships are being recorded at the level of the study but not yet at the level of the map.
Why that gap is smaller than it looks: for each of those studies the table below states the claim that was actually tested. A no-effect link is only created when the test was of a link between two nodes of the map; a null result about a compound that is not on the map, or about a clinical outcome, is recorded here instead.
| Study | Claim tested | Model | Edge | Why |
|---|---|---|---|---|
| MAN2021 | Low-dose TORC1 inhibition with RTB101 reduces clinically symptomatic respiratory illness in adults aged 65 and older. | Phase 2b and phase 3 randomised trials in older adults (phase 3 n=1024) | EVE-IMMUNE (conflicting study) | Recorded as conflicting evidence on EVE-IMMUNE. A separate no-effect edge would misstate the tested link: the trial tested a drug (RTB101, not rapamycin-like) against a clinical outcome, not an interaction between two pathway nodes. |
| STR2012 | Resveratrol, green tea extract, curcumin, oxaloacetic acid or medium-chain triglyceride oil extends lifespan in genetically heterogeneous mice. | Mouse, genetically heterogeneous (Interventions Testing Program, three sites) | none | None of the tested compounds is a node of the pathway map, so there is no edge that could carry a no-effect sign. |
| MIL2011 | Resveratrol or simvastatin extends lifespan in genetically heterogeneous mice (negative arms of a study in which rapamycin extended lifespan). | Mouse, genetically heterogeneous (Interventions Testing Program, three sites) | none | The negative arms concern compounds that are not nodes of the pathway map. The positive rapamycin arm of the same study is evidence for lifespan edges, not a null result. |
| POU2013 | Four weeks of high-dose resveratrol improves insulin sensitivity, substrate metabolism or body composition in obese men. | Humans, randomised placebo-controlled trial, obese men (n=24) | none | Resveratrol is not a node of the pathway map, and the tested link (compound to metabolic outcome in people) is not a step between two mapped nodes. |
| JIB2026 | Rapamycin co-exposure protects a mouse spermatogonial cell line (GC6-spg) from cisplatin-induced damage. | GC6-spg spermatogonial cell line, in vitro | none | One cell line and a cell-damage endpoint, not a pathway step. It limits a protective claim made in female fertility models; it is not a tested-and-null pathway link. |
| SIL2018 | Long-term daily rapamycin causes hematological or clinical laboratory side effects in middle-aged common marmosets. | Common marmoset (Callithrix jacchus), middle-aged, daily oral rapamycin for 9 months | none | A tolerability result for a drug in one primate species, not a tested link between two mapped pathway nodes, so there is no edge that could carry a no-effect sign. |
| HOR2021 | Rapamycin slows epigenetic (DNA methylation) age in the blood of common marmosets. | Common marmoset (Callithrix jacchus), whole-blood DNA methylation, rapamycin-treated cohort; clocks built with array data | none | A null result on one biomarker (blood DNA methylation age), not a pathway step. It does not show that rapamycin has no effect on aging in marmosets, only that these clocks did not register one, so it is not filed as a no-effect edge. |
| STE2026 | Rapamycin prescribed as maintenance immunosuppression is associated with longer survival in kidney transplant recipients. | Human, US national organ transplant registry: 11,474 recipients alive with a functioning graft at one year (4,885 prescribed rapamycin, 6,589 not), median follow-up 11 years, propensity-score weighting on 18 factors | RAPA-LONGEVITY (conflicting study) | Recorded as conflicting evidence on RAPA-LONGEVITY, not as a no-effect edge. The edge's support is mouse lifespan at low or intermittent doses in healthy animals; this is human observational data at immunosuppressive doses, in people on other immunosuppressants who entered the cohort by surviving twelve months after transplant. It bounds how far the mouse result reaches into people, and a no-effect sign would read as a tested-and-null pathway step, which a confounded registry comparison cannot establish. |
Who has checked this
Each link in the map has a curation state. Proposed means it was drafted and cited but has not been signed off by a second reader; Contested means the Atlas records the direction, mechanism or scope of the link as disputed, which is a statement about the literature, not about review; Confirmed means a reviewer has checked the claim against the cited papers.
contested flag and its contested_relations count come from a different field, Consensus, which records the state of agreement in the literature. A link can be awaiting review while the underlying biology is settled, and it can be disputed in the field while nobody here has reviewed it, so the two numbers are not expected to match and neither is the number of disputed claims on its own. Where they disagree for a specific link, the reason is written on that link's own page rather than averaged away -- GLN-RAG is the current case.| State | Links | Share |
|---|---|---|
| Proposed | 130 | 94 % |
| Contested | 8 | 6 % |
Checked by a researcher outside the project: 0 of 138 links. When a scientist whose own work supports a link checks it, their name is listed here and on the link's page, only with their permission.
That zero means no link currently carries a recorded external sign-off. It does not mean the map has never been read by anyone outside the project. On 29 July 2026 an external scientific review went through the 93 links the map held at that date, sign by sign, and reported no incorrect signs; its other findings were corrected and are listed in the method changelog. That pass was never written back onto the individual links as sign-off, which is why it is absent from the table. The map now holds 138 links, so at least 45 of them post-date that review entirely, and the links it did cover have been edited since. Read it as a historical check on an earlier state of the map, not as current verification of these claims.
What this audit does not measure
- It measures the corpus, not the literature. The Atlas is a hand-picked subset. Every proportion here is conditional on what one curator selected, and selection is the largest uncertainty on this page.
- It says nothing about quality. An evidence code is a statement about the kind of claim a study can support. A badly run trial still codes as human evidence; an excellent structure still codes as molecular.
- It does not weight by size or replication. A trial of twenty people and one of a thousand count the same here.
- It cannot see what was never published. Negative results that never reached a journal are invisible to any audit of the published record, including this one.
Method
Figures are recomputed on every build by build_evidence_audit.py from two sources: the study corpus (atlas_data/studies_baked.json) and the pathway links baked into the map. No number on this page is typed by hand, so a change in the data changes the page and nothing else has to be remembered.
Link grading takes the strongest evidence code among the studies citing that link. Human evidence means codes S and H. Period counts use five-year windows by publication year; undated records (1) are excluded from that table only.
The audit is written to be pathway-agnostic: the same script, pointed at a different corpus and map, produces the same measurements for any signalling pathway. The interesting comparison is not this page in isolation but two of them side by side.
Data and code are CC BY 4.0 — see Data & Citation. Corrections are welcome and are logged; see About & Methodology.
The write-up of this audit is a preprint: Ten percent human: an evidence-graded audit of the mTOR literature and the pathway's translational gap (PDF, DOI, how to cite). It is a fixed snapshot; the numbers on this page are recomputed on every build and will drift from it as the corpus grows.
Thank you. Your answer was counted anonymously.