DDX41

Summary: DDX41 participates in spliceosomal mRNA splicing.

Reason: The target leaf PTN002776112 descends from the DDX41-related node PTN002776089, where the current IBD places spliceosomal splicing. Human Q9UJV9 is a legitimate experimental descendant source, consistent with the original C-complex proteomics and later target-specific splicing perturbation/CLIP studies. The self-source is not circular, and the established clade placement remains sound.

Summary: C-complex membership supports the existing curator inference to splicing.

Reason: The original PMID:11991638 Methods and Table 1 are accessible as indexed original text: native complexes were assembled in HeLa nuclear extract on a substrate arrested after the first catalytic step, and Abstrakt/Q9UJV9 was identified among proteins found only in C complex. This supports the IC from GO:0071013. It does not identify DDX41 as the enzyme that cleaves the splice sites.

Summary: The DDX41-family mapping to spliceosomal splicing is supported.

Reason: The literal InterPro2GO rule for IPR044113 assigns GO:0000398. Its target-family context is independently supported by human C-complex identification and the later splicing studies; accepting this process does not require assuming that the DEAD-box domain proves RNA-duplex unwinding.

Summary: DDX41 participates in spliceosomal pre-mRNA processing in human cells.

Reason: PMID:25920683 links DDX41 depletion in human K562 and primary CD34+ cells to increased skipping of ZMYM2 exon 3, with the reciprocal decrease after wild-type DDX41 expression in U937 cells. The same study identifies RNA-resistant associations with spliceosomal proteins in human HEK293 nuclear extracts. PMID:36229594 provides stronger mechanistic support: tagged DDX41 binds coding RNA near 5′ splice sites in K562 cells and associates with activated spliceosomes and the transcription machinery. Together these observations support direct participation in spliceosomal RNA processing. They do not establish that DDX41 catalyzes RNA-duplex unwinding or the spliceosomal transesterification reaction.

Summary: Nucleic-acid binding can be resolved into supported DNA and RNA binding.

Reason: IPR011545 supplies the broad nucleic-acid-binding mapping. Target biochemical DNA binding/unwinding and independently reviewed RNA CLIP evidence support GO:0003677 and GO:0003723, respectively. These specific functions are already represented by other source annotations; the refinement adds precision to this row without adding a new source assertion or asserting that both substrates are unwound.

Proposed replacements: DNA binding RNA binding

Summary: DNA binding is supported beyond the ortholog inference.

Reason: The current mouse Q91VN6 donor has an experimental DNA-binding annotation from PMID:21892174, and its literal identity agrees with the ortholog source. The original study includes purified DDX41 DNA-binding experiments, while direct human DNA-substrate assays in PMID:35613581 independently support the target function. The older construct species is not inferred merely from its expression host.

Summary: RNA binding is supported by the capture experiment and independent target CLIP.

Reason: The recovered PMID:22658674 original and supplementary methods describe conventional/PAR UV crosslinking in HeLa cells, stringent oligo(dT) capture, RNase release and mass spectrometry. The separate protein table was not recovered, so target assignment is retained from the curated HDA row rather than claimed as a personally verified table entry. Independent DDX41 CLIP evidence establishes the same broad RNA-binding function. Neither this capture method nor the later intronic CLIP requires a mature-mRNA-only or RNA-helicase annotation.

Summary: RNA binding is supported independently; the original capture assignment is retained with curator deference.

Reason: The PMID:22681889 abstract describes photoreactive-nucleotide-enhanced UV crosslinking, oligo(dT) purification and proteomics in a human embryonic kidney cell line. Its target table and full methods remain inaccessible. Independent human DDX41 UV-CLIP in PMID:36229594 establishes RNA binding, so the same gene-level claim is accepted here while deferring to the curator for this particular HDA assignment. This does not claim that the missing target table was independently checked.

Summary: The RNA-helicase ancestral inference is credible but target retention is unresolved.

Reason: The actual PAINT tree places DDX41 below the RNA-helicase IBD node PTN002774595; experimentally grounded EIF4A3 and DDX3X donor records remain present. Thus neither donor count nor the target appearing among descendants invalidates the inference. However, the target source PMID:35613581 fails to unwind the tested 3-prime- and 5-prime-tailed RNA duplexes while unwinding DNA, and the reviewed nuclear studies establish RNA association/splicing without measuring RNA-duplex unwinding. These results leave substrate or cofactor requirements unresolved rather than demonstrating global loss of ancestral RNA-helicase activity.

Summary: The cited experiments demonstrate DNA unwinding, rather than RNA-duplex unwinding.

Reason: The original Results and Figure 3 of PMID:35613581 explicitly report no unwinding of the tested 3-prime- or 5-prime-tailed 13-bp RNA duplexes by recombinant DDX41, with active DDX43 as a positive control. In contrast, DDX41 unwinds forked and tailed DNA duplexes in an ATP-dependent manner; nonhydrolyzable ATP analogues and the ATPase-defective E345A mutant do not support this reaction. Replace this source-specific RNA-helicase assertion with DNA helicase activity. This conclusion follows the actual positive and negative substrate experiments, not the protein family name, and does not establish that DDX41 can never remodel an RNA substrate under other conditions. The reported preference for a 3-prime overhang is retained in notes without asserting a directly measured processive translocation mechanism.

Proposed replacements: DNA helicase activity

Summary: The electronic RNA-specific activity assignment remains unresolved.

Reason: The literal IPR014014 and IPR044113 rules map to RNA helicase activity, and the composite row also cites EC:3.6.4.13. Domain/family recognition is valid, but it does not settle RNA substrate specificity: PMID:35613581 demonstrates DNA unwinding and negative results with its tested RNA duplexes. Nuclear RNA-binding and splicing evidence does not itself measure unwinding. Preserve the source mapping while leaving the RNA-specific activity undecided; the direct DNA-helicase assertion is handled separately.

Summary: Retain the phylogenetic mRNA-binding inference with its mature-RNA scope.

Reason: DDX41 descends from the actual mRNA-binding IBD node PTN002776867, whose current donor records include experimental mRNA-binding evidence on EIF4A3 and DDX3X. No target-specific loss of this binding capacity has been demonstrated. Retain the considered ancestral inference as non-core. GO:0003729 excludes introns, so DDX41 binding at intron-containing splice sites is not used as direct proof of this narrower term; the nuclear core uses broad RNA binding.

Summary: The positive NKAP interaction supports transcription corepressor binding.

Reason: The literal human DDX41–NKAP pair is represented by two-hybrid and anti-tag coimmunoprecipitation records in PMID:22365833 (MINT-8337673 and MINT-8331403). Human NKAP/Q8N5F7 is an experimentally curated transcriptional corepressor. GO:0001222 describes binding to that partner class, so it is more informative than generic protein binding. The spliceosome interaction study does not show DDX41 repressing transcription or prove that its NKAP association acts in a transcriptional complex. The human NKAP partner-class record and its exact UniProt quotation are documented in the source guide; GO:0001222 specifies the bound protein class, not a transcriptional activity of DDX41.

Proposed replacements: transcription corepressor binding

Summary: The SIAH1 pair supports ubiquitin protein ligase binding.

Reason: PMID:25416956 has matching human DDX41–SIAH1 records across pooling, array and validated two-hybrid stages, including EBI-10581902. The original methods require sequence confirmation and at least three positive pairwise retests out of four. SIAH1/Q8IUQ4 is a RING-type E3 ubiquitin ligase, satisfying the partner class in GO:0031625. This is qualitative binding capacity in the assay, without evidence that DDX41 is a SIAH1 substrate, ligase regulator or component of an endogenous ubiquitination complex.

Proposed replacements: ubiquitin protein ligase binding

Summary: Preserve the validated CEP70 interaction as non-core binding.

Reason: The human DDX41–CEP70 pair has pooling, array and validated two-hybrid records in PMID:25416956, including EBI-10592580. The study's identity-confirmed repeated testing supports a positive physical association. CEP70 is a centrosomal microtubule-organization protein, but this pair does not establish DDX41 as a microtubule-binding protein, centrosomal component or assembly factor. No justified more informative molecular-function class was identified.

Summary: The ERCC6/CSB association supports ATPase binding.

Reason: The original PMID:26030138 reports DDX41 in the CSB interaction analysis and confirms association in tagged human-cell lysates by anti-Myc coimmunoprecipitation; the reciprocal FLAG experiment did not detect the pair. ERCC6/Q03468 has experimentally established DNA-dependent ATPase activity, making GO:0051117 the supported partner-class refinement. The evidence is a cellular association rather than purified binary binding, and does not establish DDX41 as a CSB activator or nucleotide-excision-repair catalyst.

Proposed replacements: ATPase binding

Summary: The NKAP AP-MS association supports the same partner-class refinement.

Reason: The literal PMID:28514442 record EBI-21520423 identifies human NKAP–DDX41 in HEK293T cells, with a C-terminal tandem tag on NKAP. The original BioPlex methods use affinity purification and mass spectrometry with background filtering. NKAP/Q8N5F7 is a transcriptional corepressor, supporting GO:0001222 as a binding-class description. AP-MS can recover co-complex association; this row does not establish direct stoichiometry or DDX41-mediated transcriptional repression. The human NKAP partner-class record and its exact UniProt quotation are documented in the source guide; GO:0001222 specifies the bound protein class, not a transcriptional activity of DDX41.

Proposed replacements: transcription corepressor binding

Summary: Preserve the validated NKAPL interaction without transferring a repression mechanism.

Reason: The human DDX41–NKAPL pair has five HuRI records covering two tag configurations and pooling/array/validated stages, including EBI-24640582. The original study uses quadruplicate pairwise retesting and sequence confirmation. NKAPL/Q5M9Q1 has a transcription-repressor function assigned by similarity to mouse NKAPL, rather than the directly supported human NKAP corepressor characterization. Retain the positive pair as non-core binding without inferring DDX41 transcriptional repression or treating NKAPL as experimentally interchangeable with NKAP.

Summary: Preserve the validated AIDA interaction as non-core binding.

Reason: The human DDX41–AIDA pair has pooling, array and validated HuRI records, including EBI-25270277; tag placement and target/partner identities are recorded explicitly. These support assay binding capacity. AIDA's inferred regulatory roles do not supply an established catalytic or adaptor activity for DDX41, and the pair lacks a demonstrated physiological mechanism that warrants a more informative molecular-function annotation.

Summary: Preserve the FAM50A association as a spliceosome-related interaction.

Reason: The original PMID:32703943 uses sequence-confirmed full-length V5-tagged human DDX41 and detects association with endogenous FAM50A in U-87 MG cells with controls. This is actual target evidence even though no matching pair was returned by the fresh MITAB queries. FAM50A is a spliceosome-associated protein, but the coimmunoprecipitation does not establish a catalytic or bridging activity for DDX41. Retain generic binding as non-core.

Summary: Preserve the HTT-fragment interaction with explicit construct limits.

Reason: The six human DDX41–HTT records in PMID:32814053 cover pooling, array and validated two-hybrid stages for HTT regions 1–511 and 1932–2642. The recovered original methods test reporter autoactivation and use multiple interaction reporters; the pair records preserve the specific fragment and polyglutamine-feature context. A canonical sequence checksum identifies the source protein, not a full-length experimental HTT construct. This supports non-core assay binding without a demonstrated native neuronal complex or DDX41 role in huntingtin aggregation.

Summary: NKAP associations in two cell lines support corepressor-partner binding.

Reason: The literal PMID:33961781 records EBI-54702916 and EBI-54932614 identify DDX41–NKAP associations in HEK293T and HCT116 cells, respectively, with the tagged bait orientation preserved for each. The original study uses independently filtered AP-MS networks and sequence-verified C-terminal FLAG-HA constructs. NKAP's established corepressor class supports GO:0001222, while co-purification does not show that DDX41 represses transcription or forms a purified binary complex. The human NKAP partner-class record and its exact UniProt quotation are documented in the source guide; GO:0001222 specifies the bound protein class, not a transcriptional activity of DDX41.

Proposed replacements: transcription corepressor binding

Summary: ATP binding is supported by the mapped domain and target biochemistry.

Reason: The literal IPR011545 and IPR044113 rules map to ATP binding. Direct DDX41 ATP hydrolysis and ATP-dependent DNA unwinding establish ATP utilization independently of the uncertain RNA substrate assignment. This nucleotide-binding annotation does not specify the nucleic-acid substrate or imply an additional independent core function.

Summary: DDX41 is predominantly nuclear before DNA stimulation.

Reason: PMID:35613581 examines endogenous DDX41 as well as GFP-tagged full-length protein in human cells. Endogenous protein is predominantly nuclear, and full-length GFP-DDX41 in HeLa cells is predominantly nuclear before DNA stimulation. Redistribution to the cytoplasm after stimulation is compatible with this nuclear annotation. The engineered 127-622 NLS-deletion construct is not evidence for an independently established natural isoform.

Summary: Full-length human DDX41 localizes predominantly to the nucleus.

Reason: PMID:26712909 reports that expressed wild-type DDX41 and the full-length R525H and R164W proteins are predominantly nuclear in human embryonic kidney cells. The approximately 55-kDa product favored by the M1I start-loss construct shows reduced nuclear and increased cytoplasmic localization; this does not contradict nuclear localization of the full-length protein. The study suggests alternative initiation at M132 or M155 but does not resolve the start site. Nuclear localization is consistent with DDX41 participation in spliceosomal RNA processing, without implying that all DDX41 products or cellular contexts are exclusively nuclear.

Summary: The nuclear mapping can be refined to supported nucleoplasmic localization.

Reason: The composite electronic source includes the nuclear ARBA rule, mouse ortholog and UniProt subcellular-location mapping. Independent target HPA immunofluorescence and the target's spliceosomal localization support nucleoplasm as a more informative location. GO:0005654 is already represented by existing source rows; the refinement does not assert that all DDX41 is confined to nucleoplasm or that the broad electronic mapping itself resolved the subcompartment.

Proposed replacements: nucleoplasm

Summary: Nucleoplasmic localization is directly supported by HPA.

Reason: The official target HPA record reports nucleoplasm as the main localization, with nucleoli as an additional localization, from antibody-based immunofluorescence in the recorded human cell lines. This supports the nucleoplasm annotation without treating the compartment as exclusive or inferring a new molecular function from staining.

Summary: The Reactome C-star assembly source places DDX41 in nucleoplasm.

Reason: R-HSA-9770141 directly includes DDX41 entity R-HSA-9769601 as an input and within output C-star complex R-HSA-9770118. The entity's reference is Q9UJV9 and its compartment is nucleoplasm. This is a specific target path rather than generic pathway membership; it supports localization without assigning the ATP-driven remodeling reaction to DDX41.

Summary: The Reactome exon-ligation source contains nucleoplasmic DDX41.

Reason: R-HSA-9770236 connects C-star complex R-HSA-9770118 to P complex R-HSA-9769587; both explicitly contain DDX41 entity R-HSA-9769601, which resolves to Q9UJV9 in nucleoplasm. The location is supported. DDX41's inclusion does not mean that it supplies the catalytic RNA/metal center responsible for exon ligation.

Summary: The Reactome P-complex release source places DDX41 in nucleoplasm.

Reason: R-HSA-9770847 has the DDX41-containing P complex as input and explicitly releases DDX41 entity R-HSA-9769601 among its outputs. That entity resolves to human Q9UJV9 in nucleoplasm. The localization is therefore directly traceable, without assigning DHX8/DHX15 remodeling chemistry to DDX41.

Summary: DDX41 associates with spliceosomal complexes in human nuclear extracts.

Reason: PMID:25920683 immunoprecipitated V5-tagged wild-type and R525H DDX41 from human HEK293 nuclear extracts and identified associated proteins by mass spectrometry. PRPF8 and SF3B1 associations were supported by immunoblotting, and spliceosomal protein associations persisted after RNA nuclease treatment. PMID:36229594 independently found nuclear spliceosome associations in K562 cells and interactions with PRP19, CDC5L and stage-associated factors in HEK293FT cells, consistent with activated spliceosomes. This supports the imported general spliceosomal-complex annotation. It does not establish direct pairwise binding to every recovered subunit or an independently demonstrated RNA helicase activity.

Summary: DDX41 can localize in the cytoplasm in an immune-signaling context.

Reason: PMID:23222971 reports cytoplasmic colocalization of expressed Myc-DDX41 and HA-TRIM21 in HEK293T cells, with the association changing after DNA stimulation. This supports the broad cytoplasmic annotation in the tested expression context. It does not establish exclusive cytoplasmic residence, endogenous localization in every cell type, or a separately resolved soluble cytosolic compartment. Later endogenous localization and fractionation experiments in PMID:35613581 independently support stimulus-dependent nuclear-to-cytoplasmic redistribution.

Summary: The cytoplasmic location is independently supported; this experimental source is retained with its access caveat.

Reason: The PMID:25704810 abstract supports BTK-dependent DDX41 phosphorylation and DNA/STING interactions, but the original localization assay remains inaccessible. Cytoplasmic localization is independently supported by the human-cell observations in PMID:23222971 and PMID:35613581. Accept that established gene-level claim and defer to the original curator for this source-specific EXP assignment. The unrecovered paper does not itself justify a more specific compartment or a claim of exclusive cytoplasmic residence.

Summary: The cited study resolves a stimulus-dependent cytosolic DDX41 pool.

Reason: PMID:35613581 reports nuclear-to-cytoplasmic redistribution after DNA stimulation and directly examines nuclear and cytosol fractions of THP-1 macrophages in Figure 4M. Cytosolic DDX41 increases after DNA challenge; the study also examines cytosolic DNA-associated DDX41 complexes. Refine the broad cytoplasm term to the independently represented cytosol annotation from the same study. This does not imply that all DDX41 is constitutively cytosolic or that the NLS-deletion construct is a natural product.

Proposed replacements: cytosol

Summary: The broad electronic cytoplasm location can be refined to cytosol.

Reason: ARBA00026971 and UniProt SL-0086 supply the broad mapping. Independent human DDX41 fractionation in PMID:35613581 and the explicit Reactome cytosolic entities support GO:0005829. Cytosol is already represented by existing source rows. The replacement adds subcompartment precision while retaining the source mapping and does not exclude conditional association with membranes or the nucleus.

Proposed replacements: cytosol

Summary: Retain context-dependent ER association from the ortholog source.

Reason: The current mouse Q91VN6 donor carries experimental ER localization from PMID:21892174. The original describes endogenous mouse D2SC fractionation and tagged DDX41 localization in human HEK293T cells, with redistribution after DNA stimulation. This makes conditional ER association defensible, while construct-species details are not inferred from the host cell and a constitutive integral-membrane or luminal role is not claimed. The principal nuclear/cytosolic functions are represented separately. ORGANELLES ER check (2026-11-05): DDX41 is a soluble DEAD-box helicase with no signal peptide or TM segment; the source shows it co-staining with STING and the ER and co-fractionating with microsome, MAM and mitochondrial fractions while unstimulated, and the authors read this as a cytosolic DDX41-STING complex that leaves the ER for endosomes after DNA stimulation. This is a recruited pool on the cytosolic face of the ER, bound to the ER-membrane adaptor STING, so the row stays at the root as KEEP_AS_NON_CORE (as for AMBRA1 and BRSK2 in ER batch 1); no ER membrane or MAM row is added, since recruitment to an organelle surface is not a membrane location.

Summary: A cytosolic DDX41 pool participates in DNA-triggered signaling.

Reason: PMID:35613581 combines microscopy, nuclear/cytosol fractionation in THP-1 macrophages, and cytosolic DNA-associated complex recovery to support a cytosolic pool after DNA stimulation. Endogenous DDX41 association with cGAS increases in the dsDNA-stimulated condition. The formaldehyde-crosslinked DNA-complex assay does not by itself prove direct molecular contact with every nucleic acid recovered; purified-protein DNA experiments provide separate support for binding and remodeling.

Summary: The viral-DNA sensing event explicitly uses cytosolic DDX41.

Reason: R-HSA-3134822 includes DDX41 entity R-HSA-3134807 as an input, resolving directly to human Q9UJV9 in cytosol. The source models association with a viral nucleic-acid ligand; original ligand and target evidence was reviewed separately. This location does not imply that DDX41 synthesizes cGAMP.

Summary: The TRIM21 event explicitly locates its DDX41 substrate in cytosol.

Reason: R-HSA-3134946 uses cytosolic Q9UJV9 entity R-HSA-3134807 and produces a ubiquitinated DDX41 complex. This supports the location. DDX41 is the ubiquitination substrate in this event, while TRIM21 supplies the E3 activity; no ubiquitin-transferase activity is assigned to DDX41.

Summary: The cyclic-dinucleotide source explicitly contains cytosolic DDX41.

Reason: R-HSA-9013869 includes cytosolic DDX41 entity R-HSA-3134807 and bacterial cyclic-dinucleotide ligands. PMID:23142775 reports DDX41 binding assays and responses in mouse and human systems; some exact tagged-construct details remain in unrecovered supplementary methods. The explicit target entity supports cytosolic localization without creating an additional signaling core or transferring cGAS synthesis activity.

Summary: The historical apoptosis source does not permit a confident human-target decision.

Reason: The cached PMID:10607561 abstract explicitly studies Drosophila Abstrakt and developmental/apoptotic phenotypes, but the full original could not be recovered through the attempted publisher routes. The human DDX41 TAS assignment cannot be adjudicated from that abstract alone. Preserve it as undecided rather than assert a wrong-gene error or treat necessity/pleiotropic cell-survival effects as an established core mechanism.

Summary: DDX41 perturbation alters hematopoietic cell proliferation in a context-dependent manner.

Reason: PMID:25920683 reports increased proliferation after DDX41 knockdown in human K562 cells, increased colony formation and serial replating by primary human CD34+ cells, and reduced growth after DDX41 expression in U937 cells. These experiments support the existing broad proliferation annotation. The direction is not universal: biallelic Ddx41 loss or hypomorphic mutation impairs mouse hematopoietic progenitor growth, and later human THP1 and K562 knockdown experiments also suppress proliferation (PMID:34473945; PMID:36229594). Retain the experimentally supported cellular phenotype as non-core, reflecting the growth consequences of RNA-processing defects and allele, dosage and model dependence; do not infer a dedicated proliferation activity or a universal positive or negative regulatory role.

Summary: The membrane-proteomics assignment remains source-specific uncertainty.

Reason: PMID:19946888 describes membrane-enriched proteomics in human NK cells, but the target entry and complete original methods were not recovered. GO:0016020 includes proteins attached to membranes, so the absence of a transmembrane segment would not refute the annotation. Retain the original HDA source as undecided without converting a broad membrane-fraction observation into integral-membrane localization.

Summary: DDX41 hydrolyzes ATP during nucleic-acid remodeling.

Reason: PMID:35613581 directly measures ATP hydrolysis by purified DDX41 with M13 single-stranded DNA as an effector, alongside DNA unwinding and nucleotide-dependence assays. The E345A mutant lacks detectable ATP hydrolysis and DNA unwinding, and R525H has reduced ATP turnover and unwinding while retaining strand annealing. The tested tRNA and rU30 do not stimulate ATPase activity. ATP hydrolysis is part of the remodeling mechanism, not a second independent core function.

Summary: The Rhea ATP-hydrolysis mapping agrees with direct target assays.

Reason: The official RHEA:13065 reaction is ATP plus water yielding ADP, phosphate and a proton, matching GO:0016887. PMID:35613581 directly measures DDX41 ATP hydrolysis and tests catalytic mutants. Accept this biochemical activity; it does not by itself establish ATP-dependent RNA unwinding or a separate ATPase core in addition to the DNA motor.

Summary: DDX41 perturbation affects hematopoietic differentiation.

Reason: PMID:25920683 reports slightly impaired apigetrin-induced erythroid differentiation after DDX41 knockdown in human K562 cells, increased CD11b and CD14 differentiation markers after DDX41 expression in human U937 cells, and altered c-kit/Gr-1 markers in murine LSK cells after knockdown. These results support the broad existing differentiation annotation, with human and mouse observations kept distinct. The experiments establish a cellular consequence of changing DDX41 function, without identifying a separate lineage-specifying molecular mechanism. Retain this developmental outcome as non-core alongside the more directly supported nuclear RNA-processing role.

Summary: The original C-complex proteomics explicitly identifies human DDX41.

Reason: Accessible indexed original Methods and Table 1 of PMID:11991638 identify Abstrakt/Q9UJV9 among proteins found only in purified C complex from HeLa nuclear extract. The substrate was arrested after the first cleavage step, matching GO:0071013. The table reports the protein assignment, not DDX41's direct catalytic participation in splice-site chemistry; no specific native isoform beyond the accession is inferred.

Summary: DDX41 participates in DNA-dependent cGAS/STING signaling.

Reason: PMID:35613581 links DDX41 loss in human THP-1 cells to reduced DNA-triggered cGAMP production, cGAS oligomerization, STING activation and downstream signaling, with supporting conditional-mouse experiments. Purified-protein assays independently show DDX41 DNA unwinding and strand annealing and modulation of cGAS product formation. Together these provide a molecular contribution beyond a downstream perturbation phenotype. cGAS, not DDX41, synthesizes cGAMP. The proposed balance between unwinding and annealing as an intracellular switch remains a mechanistic model: direct strand remodeling in cells was not visualized, and the authors acknowledge possible contributions from altered RNA processing.