Review papers with raw data transparency

• Single preimplantation embryos (various cleavage to early blastocyst stages) from reciprocal interspecific crosses (mus↔cast) to allow allele-specific TE expression analysis. Single-embryo interspecific So-Smart-seq design
• Differentiating hybrid female ES cells to model random XCI (mus/cast hybrid with Xi fixed to mus-origin due to mutated Tsix). Random XCI model in hybrid ES cells

• They build a TE-centric, allele-discriminating pipeline by aligning reads separately to de novo assembled mus and cast genomes and using relative alignment quality to assign allelic origin (then retaining only reads uniquely aligned under their criteria to both references). Allele-aware TE pipeline via parental-genome alignment

• TEs participate in imprinted XCI, with Xp TE silencing that is Xist-dependent (paternal Xist knockout disrupts TE silencing patterns). Xist dependence of Xp TE silencing
• Silencing differs by TE class and genomic location (notably SINEs show strong resistance/escapee behavior, and silencing kinetics correlate with linear distance/proximity to Xist during imprinted XCI). Class/proximity-dependent TE XCI silencing
• No Xa-hyperactivation for TEs: after normalizing relative TE expression dynamics between Xa vs autosomes (and validating their approach on gene elements), the TE relative expression on Xa is reported as statistically indistinguishable from autosomes across most stages tested. Method-validated absence of TE Xa hyperactivation

1) Allele assignment for multi-copy TE reads

• Strength: They attempt to reduce allelic misassignment by using separate parental-genome alignments and requiring uniqueness criteria; they also report validation using bulk fibroblast RNA-seq from pure parental strains, showing exclusive mapping of mus vs cast TE reads to their respective parental references. Parental-strain validation of TE allele mapping
• Key limitation (explicitly inherent): The pipeline discards many repeat-associated reads due to multi-mapping, retaining only uniquely aligned reads. This improves specificity but can introduce coverage bias across TE families/loci, because different TE copies may vary in mappability/uniqueness under the chosen reference assemblies and alignment thresholds. Uniqueness filtering reduces multi-mapping TE reads
• Consequence for interpretation: If uniquely mappable subsets are systematically enriched for particular TE ages/families or particular chromatin contexts, then apparent differences in silencing kinetics (e.g., “SINEs resist XCI”) could partly reflect which TE copies are represented among uniquely mappable reads rather than the biology of all TE copies. The paper partially mitigates this by focusing on allelic dynamics for thousands of TEs, but the uniqueness-filter issue remains a primary epistemic risk. Core dependence of TE quantification on uniqueness criterion

2) Xist dependence and “imprinted vs random” divergence

• Strength: The study’s central mechanistic lever is Xist dependence: they report TE Xp silencing that is impaired in paternal Xist knockout embryos during preimplantation stages. That aligns with Xist’s known role in XCI and supports TE silencing being part of the Xist-mediated system rather than an unrelated repression artifact. Xist required for TE silencing in imprinted XCI
• Imprinted-state nuance: They further claim that TE silencing differs by developmental timing, TE class, and linear distance to Xist; escapee SINEs/LTRs are described as clustering and sometimes aligning with escapee genes in 3D structure. Imprinted XCI: class-, distance-, and 3D-structure-associated silencing/escape
• Counterpoint risk: The “proximity-to-Xist” relationship is correlational and sensitive to how TE loci and Xist locus positions are represented in 2D genomic distance vs 3D spatial proximity. The paper itself discusses that random XCI differs, which may reflect differences in Xist spreading geometry, but the evidence provided (as described in the text) remains indirect unless accompanied by direct locus-level Xist/RNA polymerase occupancy or chromatin measurements for the specific TE escapee copies. Proximity effects are correlational and may depend on spatial assumptions

3) The headline claim: no TE Xa-hyperactivation

• Strength: The authors explicitly validate their quantification logic by checking that the same relative-expression method recovers known gene Xa hyperactivation behavior (reported as appearing starting at 8C in embryos and in ES differentiation). Method validated on gene hyperactivation benchmark
• Then they apply it to TEs and report statistical indistinguishability of TE relative expression between Xa and autosomes across most stages. That is a fairly direct “within-assay” negative result for the TE portion of the dosage compensation mechanism. TEs lack Xa hyperactivation in their assay
• Main skeptical blind spot: A negative finding is only as strong as the assay’s ability to detect changes in TE expression. If Xa hyperactivation were locus-specific, copy-class specific, or restricted to a subset of uniquely mappable TE reads, the aggregate test might miss it. The paper does not (in the excerpt) report locus-level detection power or whether the TE-read uniqueness filtering could selectively blunt hyperactivation signatures. Sensitivity constrained by uniquely mappable allele-assigned TE reads

• Allelic mapping pipeline: align-to-parent references (mus & cast de novo assemblies), read uniqueness thresholds, TE annotation source used for TE definitions, and the exact criteria for binomial significance of monoallelic vs biallelic. Allelic pipeline details: mapping, filtering, TE annotation, binomial allele calling
• Normalization logic for Xa hyperactivation: confirm the relative-expression normalization steps (normalizing to beginning stage for Xa/autosomes and then comparing dynamics). Xa hyperactivation quantification scheme
• Data availability: GEO accessions for TE expression and single-embryo So-Smart-seq; single-cell Hi-C and downstream usage. GEO accession numbers for all key datasets

• Mappability / uniqueness bias for repeats can differentially sample TE copies, possibly distorting class- or locus-specific conclusions. Uniqueness filtering introduces TE coverage bias
• Correlational inference of 3D mechanisms: clustering near Xist and colocalization in topological domains is presented as supportive but may not identify causal links between TE loci, Xist spreading, and TE transcription control. 3D proximity evidence is inferential/correlational
• Generalizability across strains: the study uses MC/CM crosses (mus/cast reciprocity) and a specific hybrid ES configuration. TE silencing differences are described as present, but broader strain diversity is not assessed in the excerpted methods summary. Limited strain systems and observed genotype effects

• Demonstrate TE Xa hyperactivation with a locus-aware measurement that is sensitive to mappability issues (e.g., evidence that specific TE loci/copied families show significant Xa-specific upregulation relative to autosomes in the same developmental windows). Falsifier: TE Xa hyperactivation detectable by better sensitivity
• Show TE silencing is not Xist-dependent (e.g., TE expression persists on Xp despite Xist disruption under comparable developmental contexts). Falsifier: TE silencing persists without Xist
• Demonstrate that escapee locus clustering near hotspots is an artifact of the allele-assignment/unique-read filtering rather than biological maintenance of transcription in Xp escape regions. Falsifier: hotspot clustering disappears with alternative TE quantification