The review preserves all 170 imported assertions and their original references. This guide distinguishes the source actually inspected from a cache's availability flag. It is an index to primary evidence and factual extracts, not an independent experimental source.
| Source | Actual evidence inspected | Scope and limitation |
|---|---|---|
| 26387594 | Complete cached Methods, Results and Discussion, independently read for the core | Three human biallelic loss-of-function cases; GAS8 and DRC3 loss, preserved DNAH5/CCDC39/DNALI1, reduced amplitude and doublet misalignment. The mouse experiment concerns CCDC40 mutants. Reported two-form antibody recognition is not sequence-resolved functional equivalence. Supplemental pixels/movies were not independently inspected. |
| 27120127 | Authentic abstract; publisher failures and repository metadata retained | Positive human axonemal disorganization, altered bending and DRC3 localization are retained with independent corroboration. Detailed full-text imaging/construct claims remain unavailable. |
| 34169321 | Reused authenticated original-source advice, Figure 4A and corresponding Methods | Tagged DRC1–DRC4 association in HEK293T. The host does not identify the ORF species or exact natural product. No purified-interface claim. |
| 37258679 | Genuine original XML and Supplementary Table 3; physical PDF page 11, header on page 10 | Human O95995/478 DRC4 is in the N-DRC inventory. Detailed scaffold geometry is algal; human models are placed into respiratory-axoneme density. No human sperm assay is inferred. |
| 10969087 | Complete authentic abstract | Explicit human GAS11 144-residue GFP fragment and microtubule association in mammalian cells. Full construct Methods unavailable; the trypanosome-centered title is not evidence of misattribution. |
| 16643277 | Indexed original Wiley Results/Methods, including microtubule/Golgi experiments and cloning | Human tagged GAS11 and human cell/airway detection are distinct from mouse recombinant Gas8 with rodent Golgi membranes and COS7/mIMCD3 localization. Normal cache remains abstract-only. No specific tethering reaction or exact current-isoform equivalence established. |
| 17366626 | Authentic abstract and literal mouse Q60779 IDA source records | Single-particle microtubule binding and GMAD/IMAD effects. Full Methods inaccessible; no purified-human reagent identity is asserted. |
| 18396146 | Authentic abstract and literal mouse small-GTPase-binding source record | Rab3B association, GTP-state preference and NIH3T3 Golgi context. Full construct Methods unavailable; no DRC4 GEF, GAP or direct exocytosis activity inferred. |
| 21659505 | Separately recovered original PMC HTML, Methods, Results and Discussion | Human GAS8 fragments 357–478/386–478 against the human SMO C terminus in Y2H. Mouse full-length or engineered Gas8 in co-IP/cellular assays, bovine GRK2, mouse knockdown and zebrafish rescue. Significant GMAD–CT enhancement is distinguished from variable full-length effects. Protected cache is abstract-only; no supplementary payload or figure-pixel read is claimed. |
| 27472056 | Actual cache Methods/Results and supplementary captions | Mouse intron-7 gene trap retains an N-terminal fusion. Motile-ciliary phenotypes are positive; tested MEF Smo accumulation and patterning are normal. Residual compensation is not established. A391V bead flow is a nonsignificant trend, not a strong rescue result. |
| 37950557 | Genuine author-deposited HAL original, selected Methods/Results | One patient among 196 MMAF cases, homozygous splice-site variant and severe sperm defects. Splicing predicted; C-terminal antibody loss permits absent or truncated protein. No natural-isoform rescue. Normal cache is abstract-only. |
| 9790751 | Complete authentic abstract; failed full-text routes preserved | Human GAS11 characterization with mouse growth-arrest context. Target-specific proliferation mechanism remains unresolved; the TAS assertion is UNDECIDED, not removed. |
| 14702039, 15489334, 15616553 | Accessible sequence-resource abstracts/text | Genomic/cDNA provenance only; no DRC4 biochemical activity or equivalence of modern products inferred. |
The target data and source selection retain all 112 IPI assertions. There are literal current IntAct matches for all 109 HuRI pairs and the KRT27 BioPlex pair. SMO and DRC1 have positive original evidence even though this current IntAct query did not return their matching records. Absence from that query is not negative biological evidence.
HuRI PMID:32296183 was read at the main generation/characterization section. Three assay configurations, repeated pairwise testing and sequence confirmation support binary interactions. Dataset-level MAPPIT/GPCA validation is not assigned to each DRC4 pair. All parent/reference ROGIDs match the corresponding official target/partner sequences. Nineteen partners have separately recorded sufficient-binding-region features; exact ranges remain visible in the data. Examples that affect class interpretation are TRIM41 residues 86–630, excluding its RING region, and MID2-2 residues 21–705, retaining its RING. Isoform suffixes are authenticated separately, not assumed equivalent from names.
BioPlex PMID:33961781 supplies KRT27 bait and DRC4 prey in a spoke-expanded anti-tag capture record. Its independently observed HuRI pair supports the bounded cytoskeletal-protein-binding refinement. The protected main cache is marked full text, but its usable content does not include a complete Methods read. The separate larger normal-fetch variant is retained only as research provenance; main bytes are not replaced.
Partner classes come from exact current UniProt identities, functions and relevant domains, interpreted against GO definitions. The replacement terms describe what DRC4 binds, not activities performed by the partner. In particular, keratin binding is not automatically binding to an assembled intermediate filament, a kinase inhibitor is not a kinase, and a complex subunit does not automatically possess the complex's catalytic activity. Positive associations without a justified narrower class remain non-core.
The final bounded class review is documented by the unmodified official definitions and their source receipt. GO:0003712 requires regulation via binding to DNA-bound transcription factors, and GO:0001221 describes the corresponding coregulator relationship. TCEA2, EAF1 and HEXIM2's verified elongation roles do not establish that class. GO:0051087 instead describes binding to a client-folding/transport chaperone. CDC37's official functional record and PMID:27339980(https://pubmed.ncbi.nlm.nih.gov/27339980/) abstract/captions and selected main introduction establish its kinase-client contact; the structural paper is partner-class evidence, not a DRC4 assay. No DRC4 folding-client mechanism is required or asserted.
The reproducible extract records all seven literal IBD joins and DRC4's 18-node target path in PTHR31543. The relevant ancestors are PTN000774884 and PTN002834877; no IRD occurs on the captured target path. Human self-descendant evidence is expected and is not circular. Mouse Q60779, fish F1QNW4 and worm Q18689 source records were examined with their actual organism boundaries; a small donor list is not treated as weak phylogenetic support. Full donor papers were not all reread where only the literal source identity/term was needed.
The official HPA page still uses GAS8. Its XML maps to O95995/current DRC4. Antibody HPA041311 records supported ICC/IF localization in A-431, U-251MG, U2OS, ASC52telo, serum-starved hTERT-RPE1, RPTEC/TERT1 and sperm contexts. Exact locations and image URLs are retained in the extract. These are record-level observations; image pixels were not reinterpreted and no isoform specificity is claimed. Cytosol is a nonexclusive human observation; the mouse-derived cytoplasm assertions are retained separately at their original resolution.
Current official products are 478 and 453 residues, with the latter deleting residues 1–25. Sequence reconstruction matches both official FASTAs. None of the source fragments, truncation mutants or antibody bands is automatically identified as the 453-residue product. The N-DRC core uses axoneme as anatomical location and axonemal nexin link as complex membership; their relation does not justify overwriting the original anatomy qualifiers.
Genuine normal caches and every preexisting main cache are preserved without hand editing. Abstract-only flags remain literal even where a separately recovered original was read. Conversely, a full-text flag is not proof that omitted Methods were inspected. Falcon and fallback failures are documented in the execution packet; no synthetic provider report is supplied.
The public extractor uses only the Python standard library and explicit source paths. It verifies hashes before deriving pair identity, fragment annotations, HPA records and PAINT descent. It does not generate curation actions or infer native interactions. The raw source URLs and hashes are public; historical exact input files can also be supplied through the documented path map. Live URLs may have changed since capture, in which case the checksum failure is intentional. No authentication keys or hidden service are needed for saved-input reproduction.
The follow-up evidence preserves factual class assertions, literal pair counts and source hashes. Existing interaction extracts remain unchanged and reproducible using their supplied parser. The newly fetched SKIL paper is abstract-only: the positive SnoN–Smad/N-CoR mechanism supports SKIL’s corepressor class, not a DRC4 assay.
GO:0140297 requires a DNA-binding transcription-factor partner; the relevant official IBA/ISA/IDA class assertions are listed individually in the evidence file. GO:0001221 concerns a transcription coregulator, with GO:0003712 specifying regulation through DNA-bound transcription factors. LMO1’s coactivator assertion and SKIL’s measured corepressor mechanism support this class. TCEA2/EAF1/HEXIM2’s elongation context remains separately assessed. ZNF629 has cis-regulatory binding IBA but no resolved specific TF MF in the inspected record. Five ZBTB26 pair records are present. ZNF250-3’s 46–555 and ZNF792’s 80–632 sufficient-binding regions retain their zinc-finger arrays; this does not establish intact KRAB repression modules or full-length assay reagents.
GO:0019900 kinase binding and GO:0019902 phosphatase binding fit PNKP’s DNA-end enzyme activities. The more specific protein-kinase/protein-phosphatase classes do not. The complete captured enzyme-binding descendant set has no suitable narrower class for GSTA4, KDM1A, EHHADH or NDOR1. The official CDC37 record and PMID:27339980(https://pubmed.ncbi.nlm.nih.gov/27339980/) abstract/structural captions support direct kinase-client contact; the current GO:0051087 definition does not require proof that DRC4 is the client. GOLGA2’s TPX2-dependent microtubule nucleation/capture and spindle/centrosome roles fit the cytoskeletal-regulatory class despite its additional Golgi functions.
For PMID:26387594(https://pubmed.ncbi.nlm.nih.gov/26387594/), inspect Methods immunofluorescence, Results DRC3/LRRC48 localization and reduced-amplitude sections, and the Discussion of conserved scaffold architecture. Human partner-retention loss is observed; improper complex assembly is inferred. No purified human reassembly assay is claimed. PMID:27472056(https://pubmed.ncbi.nlm.nih.gov/27472056/) Results/Figure 4 and Methods high-speed video/bead-flow assays support mouse epithelial cilium movement, not movement of the epithelial cell. PMID:37258679(https://pubmed.ncbi.nlm.nih.gov/37258679/) supplies algal scaffold architecture and homologous human model placement/inventory; the original human-versus-algal distinction is preserved. Together these support the two new structural/regulatory assertions without claiming human motor catalysis.
The ontology audit distinguishes is_a/part_of from regulates. Protein-containing complex assembly and axoneme assembly are not typed ancestors of one another; both can describe the same scaffold mechanism. Neither proposed NEW term duplicates a typed ancestor or descendant already present among the gene’s source/replacement annotations. Same-role comparator records for human DRC2 and algal DRC1 are included, with their actual evidence and limits. No new excerpt is reproduced in this follow-up.