Biology subjects

Zhang, C.

Publications and source records attributed to Zhang, C..

At least 91 records · Page 5Linked to original sources

bioRxiv2026

Widespread use of invalid statistical tests in biomedical machine learning

Cross-validation is routinely used to compare performance in biomedical artificial intelligence. Standard tests ignore correlation across cross-validation folds, inflating false-positive rates. In a PRISMA-guided meta-analysis of 184 studies (impact factor [≥] 15) across 30 biomedical fields, 97% use invalid tests. Among studies with abstract-level claims supported by invalid tests, 59% rely on a spurious comparison: one that loses significance after correcting for this correlation. On average, regaining significance requires a 43% larger effect. Extending these findings to the broader literature, we estimate that spurious comparisons occur in nearly two in three studies and support abstract-level claims in nearly one in three studies. Simulations confirm false-positive rates paradoxically approach 100% when cross-validation is repeated to improve stability. We introduce SHARP, a redesign of cross-validation, which best balances false-positive control and power among 13 benchmarked tests. These results reveal widespread fragility in biomedical artificial intelligence and offer a practical route to valid comparisons.

bioinformatics↗

bioRxiv2026

PlantP450Dock: an Automated Molecular Docking Pipeline of Plant Cytochrome P450s

Cytochrome P450 enzymes (CYPs) are the primary drivers of chemical diversification in plant secondary metabolism; however, fewer than 10% of the superfamily members have been functionally characterized. Computational docking provides a scalable strategy to prioritize candidates for experimental validation, yet prevailing workflows are poorly adapted to plant P450s because AlphaFold-predicted structures lack the essential heme cofactor and conventional flexible-residue selection relies on subjective geometric cutoffs. To address these limitations, we developed an automated pipeline--PlantP450Dock--that unifies heme cofactor implantation, molecular dynamics-based conformational sampling, data-driven flexible residue selection, and semi-flexible docking within a single integrated workflow. The heme is transferred from a crystallographic template to the AlphaFold model via a local coordinate transformation algorithm, achieving a positional deviation of less than 0.2 [A] relative to the experimentally determined CYP73A33 structure (PDB: 6VBY). Subsequent 100 ns molecular dynamics simulations confirmed faithful preservation of the Fe-S coordination geometry (2.61 {+/-} 0.08 [A]) across all trajectory frames. A singular value decomposition-based heme-plane filtering strategy objectively identified distal active-site residues for flexible treatment, eliminating user-dependent subjectivity. Cross-family validation against four phylogenetically distinct P450s (CYP73, CYP711, CYP706, and CYP701) generated catalytically competent binding poses with substrate-to-heme-iron distances of 2.8-4.4 [A] without enzyme-specific parameterization. Released as an open-source tool, this pipeline furnishes the plant science community with a standardized, reproducible computational framework to accelerate functional annotation of the largely unexplored plant P450 families.

bioinformatics↗

bioRxiv2026

Large-Scale Assessment of NF1 Single Amino Acid Variants as HLA Class I Neoantigens

Neoantigens are cancer-specific antigens arising from genomic alterations. Single Amino Acid Variants (SAAVs) represent a primary class of these neoantigens. To evaluate the therapeutic potential of Neurofibromin 1 (NF1)-derived SAAVs - given that NF1 is frequently mutated in malignant brain tumors - we prioritized the 40 NF1 SAAVs determined to be HLA-A*02:01 binders using computational prediction coupled with experimental validation. To validate these predicted neoepitopes, we employed a two-tiered experimental approach in HLA-A*02:01 homozygous U87-MG cells. We first synthesized minigene constructs encoding the predicted neoepitopes, introduced them via lentiviral transfection and confirmed their expression by mass spectrometry (MS). Subsequently, we performed endogenous validation using pan-HLA immunoprecipitation mass spectrometry (IP-MS), confirming 4 (10 neoepitopes) of the 40 candidate SAAVs. We observed a discrepancy between in silico predictions and the observed sequences. Our endogenous peptidomics further revealed conserved peptide motifs and demonstrated that peptide selection for HLA presentation is transient. While our study substantiates the therapeutic feasibility of T-cell immunotherapies targeting NF1 mutations, these results underscore a limitation in current computational prediction. Our study highlights the necessity of experimental validation to refine neoantigen prioritization strategies.

immunology↗

bioRxiv2026

Dual-view Guided Context-aware Network for Automated Bone Lesion Segmentation and Quantification in Whole-body SPECT

Whole-body SPECT bone scintigraphy reflects skeletal metabolic activity throughout the body and plays an indispensable role in the screening, treatment evaluation, and prognostic assessment of bone metastases in tumors. However, the automatic detection and segmentation of hypermetabolic bone lesions remain challenging due to low contrast, limited spatial resolution, and complex lesion distributions. In this study, we proposed Bone-Segnet, a dual-view guided automatic segmentation network for hypermetabolic bone lesions that integrated multi-scale feature modeling, global context modeling, and view-conditioned modulation. Pixel-level annotated anterior and posterior whole-body bone scintigraphy images were used for model training and prediction. The proposed network enhanced the recognition of low-contrast and small-scale lesions through small-lesion enhancement and multi-scale contextual modeling. A Transformer module was further introduced to strengthen global feature representation, while cross-view collaborative modeling was achieved by incorporating the complementary characteristics of anterior and posterior imaging. Experimental results demonstrated that the proposed method outperformed existing approaches across multiple evaluation metrics, with the Dice score improving from 0.7440 to 0.8750, indicating a substantial improvement in segmentation performance. Further quantitative analysis based on the segmentation results revealed significant differences among disease types in lesion count, pixel burden, and spatial distribution patterns, reflecting the heterogeneity of disease-related skeletal metabolic activity. Overall, the proposed method improved automatic lesion segmentation performance and enabled quantitative analysis of lesion burden and spatial distribution patterns, providing objective data support for the assessment of related diseases. Index Terms--Whole-body SPECT, bone lesion segmentation, dual-view modeling, quantitative analysis.

bioinformatics↗

bioRxiv2026

Rodent-driven NO3--N enrichment reshapes amoeba--bacteria co-occurrence and bacterial functional potential in burrow soils

Interactions between amoebae and bacteria are increasingly viewed as key drivers of zoonotic pathogen emergence in rodent-dwelling burrows, yet the environmental factors shaping these interactions remain poorly understood. Here, we analyzed soil characteristics and used absolute quantitative high-throughput sequencing to assess microbial communities in active burrow, inactive burrow, and off-burrow soils across four rodent species (marmot, squirrel, gerbil, and vole) in the Hulunbuir grassland of Inner Mongolia, China. This study demonstrates that rodent activity creates chemically distinct soil microhabitats, with nitrate (NO --N) enrichment in active burrow soils consistently observed across rodent species. Elevated soil NO3--N was associated with reduced microbial phylogenetic diversity and reorganization of amoeba-co-occurring bacterial assemblages. Both absolute abundance-based correlations and functional prediction of co-occurring bacteria indicated that amoebae were primarily associated with nitrogen-cycling bacteria in off-burrow soils. In burrow soils, amoebae increasingly interacted with bacterial taxa associated with pathogenicity while retaining ties to nitrogen-cycling taxa. Structural equation modeling and mediation analysis revealed that NO3--N enrichment indirectly linked to increased infectious disease-related functional potential by amoeba-associated bacterial restructuring and coordinated shifts in nitrogen cycling, independent of changes in bacterial abundance. Together, our findings highlight the importance of rodent-driven soil heterogeneity in shaping amoeba-bacteria interactions and suggest that rodent-mediated NO --N enrichment may promote the emergence and persistence of potentially pathogenic bacteria, with broader implications for soil ecosystem functioning and disease-related processes in terrestrial ecosystems.

ecology↗

bioRxiv2026

Dual inhibition of GTP-bound (ON) and GDP-bound (OFF) KRASG12C suppresses PI3Kα and leads to potent tumor inhibition

Current approved KRASG12C inhibitors covalently bind the inactive GDP-bound (OFF) form of KRASG12C. Recently, KRASG12C inhibitors that selectively bind to the GTP-bound (ON) form of both KRASG12C (ON) and (OFF) forms have been reported and entered clinical testing. In principle, KRASG12C (ON) inhibitors may be less susceptible to adaptive mechanisms that promote resistance to (OFF) inhibitors, however the specific mechanisms that differentiate the activity of (ON) versus (OFF) inhibition are not well understood. We profiled the activity of BBO-8520, a covalent dual inhibitor of GTP-bound (ON) and GDP-bound (OFF) KRASG12C, in KRASG12C-mutant non-small cell lung cancer models. BBO-8520 exerted more potent and sustained inhibition of KRASG12C and anti-tumor activity in vitro and in vivo compared with sotorasib, a KRASG12C (OFF)-only inhibitor. While cells treated with BBO-8520 or sotorasib both exhibited feedback reactivation of MAPK signaling driven by wild-type HRAS/NRAS isoforms, more durable suppression of KRASG12C by BBO-8520 was associated with decreased PI3K-AKT activation in vitro. Disruption of the interaction between RAS and PI3K using a novel protein:protein interaction inhibitor suppressed PI3K-AKT activation and increased the tumor response to sotorasib to a similar level as BBO-8520. Moreover, in some contexts, disruption of RAS-PI3K further increased the anti-tumor activity of BBO-8520 monotherapy. These results reveal mechanistic differences between KRAS (ON) and (OFF) inhibitors, highlight the importance of PI3K-AKT signaling in driving resistance to KRAS inhibition in lung cancer, and suggest combination strategies that suppress PI3K-AKT to improve the response to KRAS inhibitors.

cancer biology↗

bioRxiv2026

Advancing ab initio genome annotation with OrionGeno

The rapid expansion of eukaryotic genome sequencing has created an urgent demand for accurate and scalable genome annotation. Existing ab initio methods often struggle to reconstruct complex gene architectures and generalize across distant lineages, limiting their use for large-scale annotation. Here we present OrionGeno, a phylogeny-aware deep learning model for end-to-end eukaryotic genome annotation. OrionGeno integrates phylogenetic context, long-range sequence modeling and joint prediction of gene structures and repetitive elements to annotate exons, introns, untranslated regions and repeats directly from genomic sequences. Applied to chromosome-level eukaryotic genomes from NCBI that lack annotations, OrionGeno generates annotations for more than 5,300 genomes, substantially expanding public annotation resources. Across diverse eukaryotic lineages, OrionGeno outperforms state-of-the-art methods at the exon, gene, protein-sequence, and protein-structural levels. It also identifies candidate protein-coding loci absent from reference protein-coding annotations in well-curated genomes. Together with a web platform and integrated annotation database, OrionGeno provides a scalable and accessible framework for translating genome assemblies into functional biological resources and supporting large-scale biodiversity initiatives such as the Earth BioGenome Project.

bioinformatics↗

bioRxiv2026

mChIP-seq for Multiplex and Multifactorial Epigenomic Profiling Uncovers Cancer-specific Histone Features in Cellular and Circulating Nucleosomes

Epigenomic profiling facilitates access to investigate regulatory roles of histone marks in a type-specific cell, and serves as a critical path for discovering noninvasive epigenetic models in cell-free nucleosomes. Here, we present mChIP-seq, an epigenomic profiling technology that is compatible with both cell and cell-free samples for synchronously profiling multifactorial epigenetic landscapes on multiple samples. Combining sample indexing in a single reaction with a pool-and-split strategy for immunoprecipitation, mChIP-seq enhances efficiency and reduces cost. Using mChIP-seq, we profiled H2A.Z and 10 histone modifications in cell lines representing 9 cancer types. Integrative analyses further revealed an atypical association of H2A.Z and H3K4me3 at promoter regions in cancer. Based on mChIP-seq, we developed cf-mChIP-seq for circulating nucleosomes, which requires as little as 25 l of plasma per profile. Profiling 38 plasma samples for H2A.Z, H3K4me3, H3K27ac, and H3K9me3 with cf-mChIP-seq revealed distinct histone mark-associated cfDNA fragment patterns in breast cancer versus healthy control, highlighting the potential of cf-mChIP-seq to expand liquid biopsy methodologies. These results demonstrate that mChIP-seq is a widely applicable technology for large-scale epigenomic profiling of nucleosomes in cellular or cell-free forms.

genomics↗

bioRxiv2026

Chemical suppression of a bacterial immune system revives repressed phages

Many antiviral immune systems have recently been discovered in bacteria. The mechanisms of several are obscure, as are their individual significance for antiphage defense. To shed light on the mechanism and significance of the two-component type I Thoeris antiphage immune system, we leveraged high-throughput phenotypic screening to identify three small molecule inhibitors. The inhibitors target the ThsA NADase component, inhibiting its 3'-cADPR-activated filamentation. The temporal control afforded by the small-molecule inhibitors allowed us to answer an outstanding question in antiviral immunity--is persistent immunity required to repress phage titers, or do immune systems become unnecessary after eradicating infectious phages? We found that Thoeris immunity must be maintained, as chemical inhibition enabled repressed phages to revive and overtake the bacterial population. Furthermore, due to the cooperative nature of antiviral immunity, we found that Thoeris must be inhibited in only 10% of the bacteria to cause phage-induced lysis of the entire population.

biochemistry↗

bioRxiv2026

Transdiagnostic Neurobiological Biotypes of Trauma Timing: Data-driven approach of Childhood and Adulthood-Onset Trauma

BackgroundThe developmental timing of trauma exposure may critically shape neurobiological outcomes, yet distinctions between childhood-onset trauma (CT) and adulthood-onset trauma (AT) remain poorly understood. AimThis study explores whether trauma onset timing is associated with distinct resting-state functional connectivity (rsFC) pattern using data-driven approach. MethodsSeventy-seven trauma-exposed individuals (Mage=36.74 years) with post-traumatic stress disorder (PTSD), PTSD with major depressive disorder (MDD), and trauma-exposed healthy controls (TEHC) underwent resting-state fMRI. Of these participants, 15 with CT only, 17 with both CT and AT, and 47 with AT only. RsFC was calculated across the amygdala, hippocampus, nucleus accumbens (NAcc), the salience (SN), default mode (DMN), and frontoparietal networks (FPN). K-means clustering identified subgroups based on rsFC, with robustness assessed via bootstrapping, cross-validation, and replication using Gaussian Mixture Modeling. The identified clusters were compared on trauma timing, type, cumulative exposure, and clinical measures. ResultsA two-cluster solution provided the most stable fit. The two generated clusters were significantly different in CT-only prevalence (p < 0.05; Cramers V = 0.26, 95% CI). The CT cluster was marked by hyperconnectivity between amygdala-FPN, DMN-SN, NAcc-SN, and hippocampus-FPN relative to the AT cluster. Individuals with both CT and AT were evenly distributed across clusters. Clusters did not differ in PTSD or comorbid diagnoses, trauma type, or cumulative exposure. ConclusionData-driven clustering revealed distinct neurobiological profiles differentiating CT and AT. CT was associated with hyperconnectivity across salience, reward, and regulatory circuits, supporting developmental timing as a determinant of brain network organization in trauma-exposed populations.

neuroscience↗

bioRxiv2026

Cell-type-specific parallel pathways in the canonical cortical microcircuit

Information processing in the cortex depends on the integration of bottom-up and top-down signals through recurrent microcircuits spanning layers. Although the canonical microcircuit provides a framework for this integration, how these interactions are implemented at synapse resolution remains unclear. Here, we use large-volume electron microscopy reconstructions of mouse primary visual cortex to map the intralaminar and interlaminar connectivity of intratelencephalic (IT) neurons in layers 2/3 and 5. We find that layer 2/3 IT neurons formed a depth-dependent gradient of recurrent connectivity, with superficial (L2) and deeper (L3) neurons potentially forming two channels associated with top-down and bottom-up processing, respectively. These channels are preserved across layers via cell-type-specific pathways involving distinct L5 IT types, rather than collapsing into a single integrative pool. Moreover, each channel is regulated by a largely separate cohort of inhibitory interneurons, stabilizing recurrent excitation while limiting crosstalk. Together, these results reveal parallel, cell-type-specific processing streams embedded within the canonical circuit.

neuroscience↗

bioRxiv2026

Reprogramming of Lipid Metabolism by FLASH Radiotherapy Selectively Protects Radiosensitive Normal Tissues

FLASH radiotherapy, delivered at ultra-high dose rates exceeding 100 Gy/s, spares normal tissues while maintaining tumor control, yet the molecular mechanisms underlying this differential response remain poorly understood. Here we employed spatial and bulk multi-omics to investigate lipid and protein remodeling in tongue tissue and Mouse oral carcinoma 2 (MOC2) tumors at two weeks post-irradiation with FLASH or conventional dose-rate (CONV) proton radiotherapy. Bulk lipidomics revealed that CONV irradiation induced marked triglyceride (TG) depletion in tongue tissue, whereas FLASH attenuated this depletion. Spatial lipidomics using MALDI-MSI demonstrated that this TG loss was spatially restricted to minor salivary glands, identifying these radiosensitive structures as focal points of radiation-induced lipid damage. Additionally, FLASH irradiation uniquely promoted increases in membrane phospholipids and their lysophospholipid intermediates, consistent with active phospholipid turnover rather than passive damage avoidance. Proteomics revealed divergent metabolic programs: CONV activated a destructive cascade characterized by Ces1d depletion, Acox1-mediated peroxisomal {beta}-oxidation, and Acot7-driven fatty acid overflow, collectively defining a lipid droplet collapse; whereas FLASH engaged a protective program featuring Mgll suppression, Apoe-mediated triglyceride redistribution, and Lypla2-mediated lysophospholipid clearance. We propose a lipid metabolic reprogramming hypothesis in which FLASH not only minimizes acute oxidative damage but actively reprograms lipid metabolism to preserve lipid droplet stores and promote membrane remodeling. These findings provide a putative molecular foundation for the FLASH effect and support published data on the protection of normal tissues.

cancer biology↗

bioRxiv2026

Melanopsin regulates axonal translation underlying retinohypothalamic circuit assembly

Intrinsically photosensitive retinal ganglion cells (ipRGCs) influence visual system development via melanopsin before photoreceptor-mediated vision, but how melanopsin signaling contributes to ipRGC circuit assembly remains unknown. Here we show that melanopsin coordinates retinohypothalamic tract development by regulating local translation in developing ipRGC axons. Loss of melanopsin selectively disrupted local translation in axons without affecting somatic translation. The affected transcripts encoded cytoskeletal regulators, adhesion molecules, and trafficking proteins, and activity-dependent changes in translation were restricted to the period before eye-opening. Consistent with impaired axonal growth and synaptogenesis, Opn4 knockout mice showed reduced ipsilateral suprachiasmatic nucleus innervation and fewer retinohypothalamic synapses, while nanoscale synaptic molecular organization and microglial engulfment were unaffected. Reduced visual drive in Opn4 knockouts further altered developmental gene expression programs across the retina, suprachiasmatic nucleus, and lateral geniculate nucleus, with region-specific differences in expression timing. These findings identify melanopsin as a regulator of local axonal translation during early circuit development, linking sensory phototransduction to translational control mechanisms that guide retinohypothalamic tract assembly and postsynaptic target maturation.

neuroscience↗

bioRxiv2026

OneGenomeRice (OGR): A Genomic Foundation Model for Rice

The transition of genomics to a predictive intelligence discipline is driven by the advent of genomic foundation models. While substantial progress has been observed in all-life and human-centric models, plant species, particularly for the staple crops, remains hindered by a lack of models. Here we introduce OneGenome-Rice (OGR), a rice (Oryza sativa) genomic foundation model pre-trained on a genomic dataset comprising 422 high-quality genomes of cultivated and wild rice. OGR is engineered upon a Mixture of Experts (MoE) transformer architecture with 1.25-billion parameters and supports an ultra-long context window of up to 1 million base (Mb) pairs at single-nucleotide resolution. A comprehensive benchmark demonstrated that OGR significantly outperforms existing state-of-the-art multi-plant or all-life genome models in 11 categories (e.g. motif identification, sweep detection, etc). We further demonstrated the utility of OGR in several downstream applications, such as indica-japonica subspecies introgression analysis, identification of agronomy trait-associated functional loci and prediction of gene expression from DNA sequences. These results establish OGR as a promising foundational computational infrastructure for rice functional genomics and precision breeding. The OGR and its fine-tuned models, including pretrained weights, training code and the rice genomic benchmark suit, have been fully opened.

genomics↗

bioRxiv2026

Scaling Multiplex qPCR Primer Design to 1000-plex using the Degenerate Incomplete Multiplex Primer List Extension (DIMPLE) Algorithm

Massively multiplexed qPCR is primarily constrained by increasing primer dimer formation as the number of distinct primers in a single reaction increases. Previous multiplex primer design algorithms either fail to sufficiently suppress primer dimers at 100+ plex, or take exceedingly high amounts of computational resources to complete. Here, we present DIMPLE, a linear-runtime primer design algorithm that effectively generates 10,000+ primers to amplify thousands of potential amplicons in a single qPCR reaction. As one clinical demonstration of this algorithm, we designed an assay to detect 2,302 distinct KMT2A gene fusion subtypes using 204 primers in a single tube. In contrast to FISH and convention NGS approaches with 2% variant allele frequency (VAF) limit of detection, our DIMPLE qPCR assay was able to analytically detect gene fusions down to 0.05% VAF. We also constructed proof-of-concept multiplex qPCR panels for additional oncology gene fusions, multiplex pathogen detection, and DNA methylation markers. The scalability and low computational cost DIMPLE are complementary to new instrument platforms for massively multiplex qPCR readout for enabling rapid, point-of-care nucleic acid testing.

bioengineering↗

bioRxiv2026

A microbiota-derived bile acid overcomes antibiotic-induced hyporesponsiveness to immune checkpoint therapy by enhancing CD8+ T cell antitumor immunity

Gut microbiota are critical determinants of effective immune checkpoint therapy (ICT), yet the microbial mediators and host mechanisms that enhance antitumor immunity remain poorly understood. Here, we identify the microbiota-derived bile acid taurodeoxycholic acid (TDCA) as a metabolite associated with immune checkpoint therapy (ICT) response. TDCA administration alone is sufficient to overcome antibiotic-induced ICT hyporesponsiveness across multiple murine tumor models. Mechanistically, TDCA directly enhances CD8 T cell-mediated antitumor immunity, increasing cytotoxicity. These effects required signaling through the bile acid receptor TGR5. Together, these findings reveal TDCA as a gut microbial metabolite that restores ICT efficacy after antibiotic disruption by directly augmenting CD8 T cell anti-tumor activity. This work supports metabolite replacement as a therapeutic strategy to mitigate antibiotic-associated loss of cancer immunotherapy response. SignificanceTDCA is a microbiota-derived metabolite that restores immune checkpoint therapy efficacy after antibiotic disruption by directly enhancing CD8 T-cell-mediated anti-tumor immunity through bile acid receptor TGR5 signaling. Our findings suggest that supplementation with defined microbial metabolites can mitigate antibiotic-associated loss of immunotherapy response without requiring broader microbiome reconstitution.

cancer biology↗

bioRxiv2026

Catalase 2-dependent regulation of autophagy in response to carbon starvation in Arabidopsis

Reactive oxygen species (ROS) are fundamental regulators of plant development and stress acclimation. However, their roles in carbon starvation tolerance remain largely unresolved. We therefore dissected the dynamics of redox regulation in carbon-starved Arabidopsis leaves. We present data showing the presence of an early ROS wave that activates autophagy, followed by a sustained, accumulation of hydrogen peroxide (H2O2) that leads to programmed cell death. We show that catalase 2 (CAT2) constitutively binds and sequesters the autophagy initiators ATG1a and ATG13a, a process that prevents the activation of autophagy under carbon-replete conditions. The accumulation of H2O2 in response to carbon starvation leads to oxidation of the Cys-370 and Cys-413 residues on the CAT2 protein and dissociation from ATG1a/13a, thereby allowing ATG1-kinase complex assembly and autophagosome formation. The expression of CAT2-C370S or CAT2-C413S variants in the cat2-1 mutant significantly enhanced carbon starvation tolerance, because both variants retain catalase activity but fail to bind ATG1a/13a. We conclude that CAT2 functions as a redox-sensitive regulator of autophagy.

plant biology↗

bioRxiv2026

A residual-ratio framework for auditing transcriptomic gene signatures against background expression structure

BackgroundTranscriptomic gene signatures are widely used to infer pathway activity and biological mechanism from bulk cancer expression data, yet current evaluation strategies primarily emphasize internal coherence, predictive performance, or scoring robustness. A quantitative framework for assessing how much signature variation remains independent of background expression structure has been lacking. ResultsUnlike existing single-number signature-quality metrics such as Berglund uniqueness, residual-ratio auditing reports a trajectory across null-model richness: for each signature we compute the residual ratio [Formula] at progressively enriched expression-PC subspaces, together with an inverse-participation-ratio (IPR) concentration diagnostic that reports the effective number of axes absorbing each signature. Applied to a curated 17-entry benchmark, all 50 MSigDB Hallmark gene sets, and 1,181 Reactome pathways across 8 TCGA cancer types (4,462 samples), with external validation in METABRIC, the framework produces two complementary readouts. First, the curated panel is absorbed into the ExprPC50 subspace at residual ratios 18-43% below size-matched random 30-gene baselines in every cancer (curated mean r{perp} range 0.109-0.177 vs. random mean 0.182-0.288), providing the frameworks central quantitative discrimination between biologically coherent signatures and arbitrary gene combinations. Second, within the curated panel the ExprPC50 residual ratio is negatively correlated with the top-5 absorption concentration in every cancer (Spearman{rho} from -0.59 in PRAD to -0.89 in SKCM, median -0.71; all 8 significant at p < 0.05, most at p < 10-3); we report this correlation as a descriptive geometric property of the null-model coordinate system rather than as a biological law, because 1,000 random 30-gene draws projected through the same top-50 expression-PC basis reproduce the same pan-cancer median{rho} (-0.73; Supplementary Table S16), and it is robust to compositional nuisance: after rebuilding the null basis as immune-PC1 {oplus} stromal-PC1 {oplus} proliferation-PC1 plus 47 residual PCs, the per-cancer{rho} becomes more negative rather than shallower (median -0.86; Supplementary Table S17), ruling out tumor purity, immune infiltrate, and stromal fraction as drivers of the pattern. Because absorption at ExprPC50 is a geometric property of how any signature direction sits in expression-PC space, tier-level distributional structure at this operating point is not separable beyond the low-vs-upper band split: a Kruskal-Wallis omnibus is significant (p = 4.9 x 10-13), but pairwise Dunns post-hoc tests show that Tiers 1, 4, and 5 are not separable (pBH > 0.2). The trajectory shape itself is empirically bootstrap-invariant: across 200 sample-level fixed-basis bootstrap resamples of the 17 curated entries in BRCA, the mean pairwise Pearson correlation of trajectory-shape vectors is 0.999, and individual cell-level 95% bootstrap CI half-widths at B= 1,000 resamples are in the range 0.002-0.053. External replication in the METABRIC breast cancer cohort (nsamples = 1,980, microarray) showed moderate-to-strong rank-ordering concordance with TCGA-BRCA across the 17 curated entries (Spearman{rho} = 0.72 on the 17-signature ordering, 95% Fisher-z CI 0.37-0.89, p = 0.001). Under an upper-bound sensitivity analysis, 45 of 50 Hallmark gene sets and 992 of 1,181 Reactome pathways had ExprPC200 residual ratios below the mean of their size-matched random baselines--a descriptive statistic reflecting axis alignment under rich null models, not a failure rate. In causal DAG simulations (nrep = 100 replicates), a signature driven entirely by a latent confounder retained r{perp} = 0.233 at ExprPC50, numerically comparable to Tier 1 validated drivers, so a single-point residual ratio cannot adjudicate confounder-independence. The frameworks load-bearing signals are therefore the trajectory shape (statistically invariant under sample-level resampling) and the magnitude gap between the curated panel and its random 30-gene baseline (the curated-vs-random discrimination), read jointly--not the value of r{perp} at any single null-model dimensionality. O_TBL View this table: org.highwire.dtl.DTLVardef@1b56848org.highwire.dtl.DTLVardef@d18636org.highwire.dtl.DTLVardef@1c26db4org.highwire.dtl.DTLVardef@1b0620corg.highwire.dtl.DTLVardef@f507d2_HPS_FORMAT_FIGEXP M_TBL O_FLOATNOTable S16:C_FLOATNO O_TABLECAPTIONRandom-gene-set null for the ExprPC50 r{perp}-vs-c(5) correlation, and curated-panel absolute gap vs random baselines. For each of the 8 primary-analysis TCGA cancer cohorts, we drew B = 1,000 random 30-gene sets from the gene universe of the preprocessed expression matrix and computed, for each draw, the residual ratio r{perp}(k = 50) and the top-5 absorption concentration c(5) under the same top-50 sample-space PC basis used for the curated benchmark (Methods [§]Statistical analysis; reference implementation accompanies the project repository as script 35). Column "Empirical curated{rho} " repeats the 17-signature Spearman{rho} between r{perp} and c(5) reported in the main text and in Supplementary Table S10; column "Null-A{rho} (random 30-gene)" gives the Spearman{rho} across the 1,000 random-draw (r{perp}, c(5)) pairs per cancer; column "{Delta} (emp null-A)" reports the difference. For reference, column "Null-B{rho} " gives the corresponding Spearman{rho} across 1,000 iid Gaussian unit vectors h (0, IN) in sample space (a uniform random direction that does not inherit the expression covariance geometry). Rightmost columns compare the curated 17-entry panels mean r{perp} at ExprPC50 to the random 30-gene baseline mean, both in absolute units and as a percentage gap; this magnitude gap is the quantitative discrimination between curated biological signatures and arbitrary gene combinations on which the frameworks central claim rests. C_TABLECAPTION C_TBL O_TBL View this table: org.highwire.dtl.DTLVardef@d4a1d3org.highwire.dtl.DTLVardef@1cc3baaorg.highwire.dtl.DTLVardef@1611aecorg.highwire.dtl.DTLVardef@2e9c76org.highwire.dtl.DTLVardef@220711_HPS_FORMAT_FIGEXP M_TBL O_FLOATNOTable S17:C_FLOATNO O_TABLECAPTIONPurity-aware null for the curated-panel ExprPC50 r{perp}-vs-c(5) correlation. For each cancer we rebuild a rank-50 sample-space basis as follows. Columns 1-3 are the PC1 directions of: an immune-infiltrate proxy panel (CD3D, CD3E, CD4, CD8A, CD8B, CD19, CD68, PTPRC, FOXP3, IFNG); a stromal/fibroblast proxy panel (COL1A1, COL1A2, COL3A1, VIM, FN1, ACTA2, PDGFRA, PDGFRB); and the 50 proliferation markers already used in the null-model hierarchy ([§]Null model hierarchy). Columns 4-50 are the top-47 PCs of the residual expression matrix Y - QbioQ Y after QR-orthonormalization of the 3-column biological block; the full 50-column basis is re-orthonormalized by a final QR pass. The 17 curated benchmark entries are then re-scored under this purity-aware basis, and the per-cancer Spearman{rho} between r{perp} and c(5) is recomputed. Column "Standard{rho} " is the value reported in the main text (same as Supplementary Table S16, column "Empirical curated"); column "Purity-adjusted{rho} " is the value under the new basis; column "{Delta}" is the difference. This test was pre-specified as a check on whether the observed{rho} (r{perp}, c(5)) could be driven by tumor-purity, immune, or stromal composition artifacts that a standard top-50 PC basis would implicitly absorb. C_TABLECAPTION C_TBL ConclusionsResidual-ratio auditing provides an interpretable and practical framework for quantifying how much of a transcriptomic gene signatures variance remains orthogonal to a chosen background-expression model. The two statistically reliable quantities it reports are (i) the shape of the trajectory r{perp}(k) across null-model richness, which is bootstrap-invariant across sample-level resamples, and (ii) the magnitude gap between the curated panels residual ratio and size-matched random 30-gene baselines at a fixed operating point, which is 18-43% in all 8 TCGA cancers and survives a purity-aware null-model construction. The negative correlation between r{perp} and the top-5 absorption concentration c (curated-panel median{rho} = -0.71) is reproduced by random 30-gene sets under the same basis (random-draw median{rho} = -0.73) and is therefore best read as a descriptive geometric property of the null-model coordinate system rather than a biological discovery about curated signatures. Any single operating-point residual ratio carries materially wider cell-level uncertainty than the trajectory shape and cannot, on its own, adjudicate confounder-independence. The frameworks outputs describe a signatures geometric relationship to modeled background expression structure and do not evaluate clinical utility: a signature with a low residual ratio may still be clinically valuable when that low value reflects alignment with a strong prognostic or actionable program such as proliferation, immune infiltration, or cell cycle, and the framework is not a substitute for calibrated prognostic or predictive classifiers. All findings are based on bulk RNA-seq (TCGA PanCancer Atlas, 8 cancer types) and microarray (METABRIC) data; transfer to single-cell, single-nucleus, or spatial transcriptomics is out of scope and not claimed. Used within this scope--reading the trajectory shape and the magnitude-gap signal jointly, rather than the value of r{perp} at any one k--the framework adds a complementary audit layer to existing pathway-scoring and experimental-validation workflows, and supports more calibrated interpretation, comparison, and reporting of transcriptomic gene signatures in cancer studies.

bioinformatics↗