ATM manual review notes

ATM executes protein serine/threonine phosphorylation in nuclear DNA-damage signaling, coordinating checkpoints and repair. It also acts in telomere regulation and extranuclear stress responses. One molecular-function core is retained; the ROS-dependent peroxisomal membrane location is explicitly contextual. There is no NEW annotation or separate lipid, generic-binding or pexophagy core.

The reviewed map after Source65 closure retains all 325 original source objects and the original null/absent alternative-products representation: 214 ACCEPT, 53 KEEP_AS_NON_CORE, 44 UNDECIDED, 12 MODIFY and 2 REMOVE. The review enums govern the known-interaction versus unresolved-source distinction. Non-core association does not mean the reported pair is false; unread source-specific pairs remain uncertain. No new molecular activity is inferred from co-IP alone, no negative evidence is inferred from a paralog-focused title, and no PAINT assertion is rejected for a short donor list or target self-inclusion. Full ancestral tree/MSA placement was not replayed.

Assay chemistry and localization corrections

Rows197–208 replace matrix with peroxisomal membrane. In PMID:26344566(https://pmc.ncbi.nlm.nih.gov/articles/PMC4589490/), selected Results and proteinase-protection Methods show accessible ATM alongside protected catalase. The correction retains association and cytosol-facing peripheral topology; it does not infer an integral membrane segment. Human fibroblast and human expression studies remain distinct from rat FAO and mouse fibroblast experiments. Cellular PEX5 phosphorylation/rescue and the in-vitro GST-p53 kinase Methods are not collapsed into purified ATM-to-PEX5 reconstitution. PEX ubiquitination, autophagy bridging and deubiquitination remain activities of the relevant other proteins. Exact nested Reactome physical-entity objects were not all replayed.

Rows246/262 remove the lipid-substrate activity and its explicit dependent GO0108 PI3P inference. The author-uploaded original PMID:11375976 Methods478–717, interaction Results775–826, kinase Results1241–1386 and Figure4D text1419–1442 were inspected independently. These are protein-substrate assays with kinase-dead controls. Wortmannin sensitivity is not evidence of phosphatidylinositol catalysis. The GO:0016303 reaction requires a lipid substrate. The complete PMID:8988033 abstract supplies a bounded negative lipid assay, not a universal absence proof. The normal11375976 cache is still abstract-only; its availability flag is unchanged by external reading. Only literal normal-cache text is quoted in the draft. No external Methods paragraph is fabricated into cached support.

The selected PMID:28508083(https://pmc.ncbi.nlm.nih.gov/articles/PMC5425235/) Results/Fig1 text and kinase Methods establish purified human ATM activity on full-length human p53 and active dimeric preparations. Historical MRN/monomer pathway representations are not universalized. MRN-linked DSB signaling and radiation-responsive p53 phosphorylation are corroborated by complete normal abstracts15790808/9733515. ATM performs the signaling step; it is not assigned catalytic activities of its substrates, modifiers or downstream effectors.

Source closure and read limits

The normal seed and69 auxiliary-source imports are closed, with all preexisting records preserved. The provider attempt failed and generated no research report; its structured closure is retained without reading or reconstructing raw diagnostics. There was no duplicate gene-wide normal source request.

Source65 recovered one abstract-only publication and eight Reactome event records, now imported without overwriting existing files. All nine exact source bodies were read, and each official event location was checked. Thirteen source-dependent rows now retain supported kinase activity or nucleoplasmic location. The NEMO association remains non-core. At shortened telomeres, the KAT5 mechanism remains explicitly an inference from DNA-break biology. The other 311 rows, one core, 12 membrane replacements and two lipid-related removals are unchanged. No new process or molecular-function assertion was added.

The owner consultation read all68 available complete cached abstracts and102 complete Reactome summaries. The independent all325 peer read every proposed rationale/source tuple, rehashed all183 evidence pins and separately verified the highest-impact chemistry/topology/core evidence. It did not newly reread every one of those sources in full. Detailed source scopes are recorded per reference below and remain attributed to that consultation. Full cached availability does not imply full-body reading; exact pair tables, all figure pixels, supplements, nested event graphs and all donor experiments remain outside the stated scope.

The combined reading covers all 69 normal cached abstracts and 110 complete Reactome summaries, with detailed original-text boundaries as attributed below. Independent scientific review passed.

Validation

On 2026-09-29, just validate human ATM passed schema, term, reference and annotation checks with 18 advisory warnings. Seventeen flag retained generic protein-binding associations; they remain non-core because source-supported interactions are not biologically false, and the evidence does not justify a more specific molecular-function replacement. One flags different actions for GO:0043065: the contextual IEA annotation is non-core, while the specific IMP source remains unresolved. Uniform actions would conceal that difference in evidence access. The review therefore remains DRAFT, with no PENDING rows. just render human ATM generated the review page. Repository-wide validation and the helper-generated history record are checked separately.

Reference-specific consultation boundaries

PR #3547 evidence follow-up — 2026-09-29

The follow-up distinguishes a cached full-text availability flag from the sections actually extracted. The normal records for PMID:29656893 and PMID:35512704 include Introduction and Discussion but omit Results, Methods and the pair tables. Their ATM interaction assertions therefore remain unresolved after reading the available body text. For HuRI (PMID:32296183), the selected Results and assay-validation text describe binary Y2H screening, while the Discussion notes one tested isoform per gene; the ATM–ESRRG-3 construct mapping remains uninspected. No publication cache was edited.

Three uncertain annotations are now accepted. GO:0071481 (PMID:26323318) is directly supported by ATM-dependent NUCKS1 Ser54 phosphorylation after 4 Gy X-rays in HeLa cells, with ATM and DNA-PK inhibitor controls. GO:1990391 (PMID:24217620) is retained by curator deference with the explicit ATM ChIP enrichment at I-PpoI-induced rDNA breaks; the ChIP experiment establishes damage-site recruitment, not physical membership or composition of a protein complex. Retention explicitly defers to the curator and the established ATM repair context, without identifying a stable ATM–LSD1 complex. GO:0007095 (PMID:15149599) is accepted as established ATM G2 checkpoint signaling with corroborating Che-1/checkpoint evidence (PMID:17157788); the original telomere paper’s detailed G2 experiments remain unread. The separate spindle-assembly-checkpoint assertion remains uncertain.

Five generic-binding records now propose GO:0004674 protein serine/threonine kinase activity because their own sources contain substrate-phosphorylation evidence: Pin2/TRF1 (PMID:11375976), DDX1 (PMID:18710941), p53R2/RRM2B (PMID:19015526), FBXO31 (PMID:19412162), and 53BP1 (PMID:22621922). The selected cached DDX1, FBXO31 and 53BP1 Results and kinase Methods were read; the Pin2/TRF1 assessment also uses the previously documented external reading and its normal abstract. The p53R2 finding is explicit in the complete cached abstract. Original IPI evidence fields and partners remain unchanged; implementing the proposed activity requires the evidence appropriate to each kinase experiment. ATM is not assigned any substrate’s enzymatic function.

The UFL1 study (PMID:30886146) establishes ATM-dependent cellular phosphorylation and positive feedback, but the selected Results, Figure 6 caption and IP Methods do not resolve its separate ATM–UFL1 physical-binding assertion. No purified ATM–UFL1 assay is claimed. The remaining supported generic associations include ATM as a TEL2 chaperone client or regulated kinase; neither role warrants inventing a more specific activity merely to avoid a generic-binding warning. Those 13 records remain non-core under the user action definitions, and the policy difference remains unresolved.

Reference relevance is now differentiated for contextual maturation, broad screens and the unresolved ATR/UPF1 paper. The TEL2 paper (PMID:20801936) directly includes mammalian ATM maturation despite also determining a yeast structure. Selected source-specific correctness assessments are now explicit; VERIFICATION of an experiment does not imply the entire paper or its figure pixels were inspected. The review contains 69 PMID references, alongside Reactome and other references; it does not contain 179 PMID records.

All 325 original annotation source objects and alternative-product representation are preserved. The current decisions are 217 ACCEPT, 49 KEEP_AS_NON_CORE, 40 UNDECIDED, 17 MODIFY and 2 REMOVE. The single kinase core, GO:0005782-to-GO:0005778 peroxisomal topology corrections, and removal of GO:0016303/GO:0036092 lipid-kinase assertions remain unchanged. The peroxisomal core location remains explicitly qualified as a documented ROS-dependent context within the same kinase activity, not a co-equal constitutive location or a separate core function. This entry supersedes the earlier decision counts; DRAFT status remains until review and policy follow-up close.

PR 3547 second follow-up: duplicate kinase replacements

The reviewer correctly identified that the Pin2/TRF1 interaction from PMID:11375976 and the FBXO31 interaction from PMID:19412162 were proposed for replacement by GO:0004674 even though each paper already supplies an accepted IDA annotation for that kinase activity. Those two proposed replacements are withdrawn. Their supported associations remain KEEP_AS_NON_CORE under the same disclosed project action convention used for the other contextual interactions. No replacement activity is now proposed from either of these two IPI assertions, and their literal evidence snippets remain intact.

The independent DDX1, p53R2 and 53BP1 kinase replacements remain unchanged. The DNA-repair-complex reason now leads with curator deference and separates the corroborating damage-site recruitment experiment from the limitation that ChIP does not establish protein-complex composition. No supporting quote, source assertion, reference judgment, product or core function changes. All 325 rows remain: 217 ACCEPT, 51 KEEP_AS_NON_CORE, 40 UNDECIDED, 15 MODIFY and two REMOVE. The two REMOVE judgments continue to concern the independently established lipid-substrate mismatch. No new annotation is added. The reviewer requested REMOVE for the two generic interactions; contextual retention instead follows the existing explicitly disclosed action convention and does not reintroduce the duplicate kinase proposals.