Bioinformatics & omics
13 articles
Genetic perturbation data lets a model forecast drug transcriptional responses in cell types never treated with the drug
- GRAFT transfers knowledge from genetic perturbation responses to chemical ones and across cellular contexts, predicting drug-induced expression changes without drug data in the target cell type.
- On LINCS and external benchmarks it outperformed competing methods, notably for context-dependent drugs.
- An atlas covering many contexts and thousands of drugs was generated, with docking support for some associations and partial support for donor-specific iPSC differences from measured genetic perturbations.
Knockdown-to-drug transfer quietly assumes a compound mimics loss of its target, which should fray for off-target or agonist-like effects. Docking agreement is also weak validation of an expression forecast. A fair test of the atlas is to hold out one cell type and run your own compounds, ideally ones with known off-target activity.
Cross-modal transfer enables zero-shot prediction of drug responses across cell types ↗
Region-level spatial gene expression beats bulk RNA signatures for ranking anticancer compounds in triple-negative breast cancer
- Bulk RNA signatures average away intratumoral differences, so the authors score compounds against localized spatial expression signatures instead, focusing on heterogeneous triple-negative breast cancer.
- They compared bulk and spatial signatures on a set of about 1,300 diverse compounds, evaluating how well each ranked known anticancer drugs.
- Spatial signatures ranked known anticancer drugs better than bulk ones and also flagged unclassified and non-oncology compounds as candidates; evidence is computational, with no experimental validation described.
Ranking known anticancer drugs higher is a retrospective enrichment test in one cancer type, and it doesn't show the spatial signature will find new ones. There's no wet-lab follow-up, and nothing on leakage or baselines. The non-oncology hits are the interesting part; I'll reserve judgment until a few are actually tested.
Blocking cGAS-STING in vessel walls curbs aortic aneurysm in mice
- Single-cell data from mouse and human aneurysm tissue place cGAS-STING signaling mainly in endothelial cells, with higher expression in ruptured than stable human lesions.
- In angiotensin II-driven mouse aneurysm, the pathway was switched on; either a cGAS inhibitor or a STING antagonist lowered aneurysm occurrence and tissue damage and improved endothelial mitochondria.
- Patient aneurysm samples also showed elevated pathway proteins, and DHFR loss emerged as a possible link between DNA damage and activation.
Two inhibitors, one mouse model: that's thin. Angiotensin II aneurysm is acute and hypertension-driven, a poor mimic of slow human AAA, and RU.521 is far more potent on mouse than human cGAS. Human evidence is correlative expression, not genetics. Next study needs dosing after aneurysm is established, exposure data, and a second model such as elastase.
CD74-ENO2 axis drives glycolytic synovial macrophages in OA; targeted siRNA nanoparticles slow disease in mice
- Single-cell and spatial transcriptomics of mouse and human osteoarthritic synovium revealed an expanding, glycolysis-reliant, proinflammatory CD74+ macrophage population at the synovium-cartilage border.
- CD74 binds the glycolytic enzyme ENO2, promoting nuclear entry of HIF-1α/β-catenin and a self-reinforcing metabolic-inflammatory program.
- Myeloid CD74 deletion or drug blockade eased synovitis and cartilage damage in mouse OA models; intra-articular lipid nanoparticles carrying CD74 siRNA, guided by MIF-mimetic peptides, reached these macrophages and slowed disease.
I would not bank on the mouse genetics alone: myeloid CD74 deletion is a clean target-validation experiment, but OA mice are mostly surgical models that skew toward inflammation more than human disease does. The harder question is delivery. A MIF-mimetic ligand on an LNP may hit CD74+ macrophages, but I want synovial clearance, off-target uptake in other CD74+ antigen-presenting cells, and knockdown duration in a human-sized joint.
CRISPR screen flags ACSL4 as a switch for obesity-linked macrophage inflammation
- A genome-wide CRISPR-Cas9 screen combined with human adipose scRNA-seq identified ACSL4 as critical for pro-inflammatory adipose tissue macrophage activation.
- Myeloid-specific Acsl4 knockout lowered inflammatory cytokine expression and protected mice from high-fat-diet obesity, with more lipolysis and thermogenesis, but glucose tolerance and insulin sensitivity did not improve significantly.
- A screen of 8,685 DrugBank compounds was run to find ACSL4-targeting candidates.
The human-data cross-check on the CRISPR hit is the right way to build a target case. But the phenotype is the problem: protection from obesity without any metabolic benefit in glucose or insulin is not what anyone wants from a metabolic-inflammation target. ACSL4 is also central to ferroptosis sensitivity, so a systemic inhibitor has an obvious safety question I'd want answered early.
Machine learning tiers PFAS liver hazard, and liver spheroid gene data back it up
- An XGBoost QSAR model trained on high-throughput bioactivity data sorts PFASs into four tiers of liver-related hazard.
- Public transcriptomes from human liver spheroids exposed to 19 PFASs showed response patterns that varied across those tiers.
- Lipid metabolism disruption and cellular stress dominated the signatures, with FABP1, PLIN2 and GDF15 responding across exposures.
Tiers from a QSAR model inherit the weaknesses of their training labels, and high-throughput bioactivity assays rarely capture slow liver injury. PFAS persistence also means spheroid concentrations may sit far from real accumulation, so checking tiers against measured tissue exposure and serum half-life matters. FABP1, PLIN2 and GDF15 look promising as readouts, but stress markers are notoriously non-specific.
Variant-guided antisense oligonucleotides selectively lower mutant ATXN3 in patient-derived neurons for spinocerebellar ataxia type 3
- Long-read nanopore sequencing of SCA3 patients from three European countries mapped many ATXN3-linked SNPs into three haplotype blocks tied to the CAG expansion.
- Modeling suggested that targeting just two SNPs from different blocks could cover most European patients with allele-specific treatment.
- Of 17 SNP-directed ASOs tested, several preferentially lowered the mutant transcript over wild-type, which retains protective deubiquitinase function.
- In two iPSC-derived cortical neuron lines, a single ASO dose strongly reduced mutant ATXN3.
Delivery will decide this more than sequence design: single-dose selectivity in two iPSC cortical lines says little about getting ASO to the cerebellum, brainstem and spinal cord at concentrations that still spare wild-type ATXN3, since allele discrimination usually erodes as dose rises. Coverage is also a European-haplotype claim; other ancestries remain untested, and no in vivo safety data are shown.
A covalent ligand at cereblon's cryptic Cys287 locks the E3 ligase open and blocks IMiD-driven degradation
- Cysteine-directed chemoproteomic screening found covalent ligands for Cys287, an allosteric pocket far from the thalidomide-binding site.
- Lead VVD-105917 broadly blocks degradation of neosubstrates driven by IMiDs and CELMoDs.
- Cryo-EM shows it traps cereblon in its open, inactive state, and ligand order matters. A previously reported ligand at the same site promotes closure instead, so the pocket is 'sign-switchable'.
Every CRBN drug so far sits in one orthosteric pocket, so a covalent handle that controls the open/closed switch is a genuinely new knob. A pan-inhibitor is an excellent tool to dissect degrader mechanisms and rescue off-target degradation. A site that can push either way also means a degrader-enhancing ligand might be hiding in the same pocket. This is a preprint, so wait for review.
Allosteric covalent inhibition of cereblon through cryptic cysteine targeting ↗
Azaphilone Probe Maps Reactive Lysines Across Live Bacterial Proteomes
- SCL-alkyne, a simplified azaphilone derived from the natural product sclerotiorin, was designed to profile lysine reactivity proteome-wide in bacteria.
- It was highly lysine-selective in vitro and in live E. coli; most engaged sites were lysines or protein N-termini, and only a small subset of reactive lysines sat in functional regions.
- Enrichment analysis pointed to bacterial chemotaxis as a major target pathway, which was then confirmed.
Covalent ligand discovery has been cysteine-centric, and lysine-directed scaffolds beyond sulfonyl fluorides and activated esters are scarce, so a natural-product-derived option is appealing. The observation that few reactive lysines are functional is a sober reminder that reactivity is cheap and function is not. As an E. coli preprint it is a starting point for covalent fragments, not a lead.
Open Screening Dataset Maps Small-Molecule Inhibitors Across Nine Human DNA Glycosylases
- A Scandinavian consortium built a high-throughput biochemical assay platform for human DNA glycosylases, the eleven enzymes that initiate base excision repair.
- A chemogenomic library was screened against nine glycosylases with assays good enough for hit calling, yielding primary and concentration-response data and several previously unreported inhibitors.
- All data, protocols and metadata are publicly available.
Glycosylases have been chemically neglected, so an open, annotated screen across nine of them is starting material a hit-finding team can mine directly. Chemogenomic libraries carry target annotations, which helps with mechanism and selectivity reads. I'd immediately cross-check the hits for assay interference and DNA-binder artifacts, since nucleic-acid enzyme screens are notorious for them.
A high-throughput screening dataset of small-molecule inhibitors across human DNA glycosylases ↗
Photocrosslinking Probes Uncover New Nonhistone Partners of the TRIM24 Bromodomain
- TRIM24 is a dual-module epigenetic "reader" whose bromodomain binds acetylated histones and is an established cancer drug-discovery target, though its nonhistone binding partners have been poorly defined.
- Researchers installed genetically encoded photocrosslinkers, AzF and BzF, at residue L922 within the bromodomain's acetyllysine pocket, preserving histone binding while enabling UV-triggered covalent capture of interacting proteins.
- Quantitative proteomics identified the nonhistone proteins NPM1 (AzF-enriched) and CDK9 (BzF-enriched), plus candidates MAP4, RAN, and SND1.
- ChIP-seq showed TRIM24, CDK9, and NPM1 co-localize near transcription start sites of cancer-associated genes; KEGG analysis tied the hits to several signaling pathways.
- The approach shows photocrosslinking can extend reader-domain interactome mapping beyond histones, yielding new candidate leads for chemical-probe and target-discovery work on TRIM24.
"Selective bromodomain inhibitor" just got a harder definition. NPM1 and CDK9 turned up riding along at L922 — the acetyllysine pocket where any TRIM24 bromodomain probe or degrader warhead would sit — so selectivity now has to mean selective against displacing these too, not just against other BRD family members. Most bromodomain campaigns are still built structure-first off that pocket alone, chasing histone-mimetic potency; this says the pocket's biology extends past histones, so cellular phenotype from a TRIM24 chemical probe could be confounded by NPM1/CDK9 disruption rather than histone reading per se. No compound here, just a crosslinker — useful, but MAP4, RAN and SND1 are "candidates" in a proteomics list, not validated partners; I'd want co-IP/FP confirmation and dose-response competition before building any SAR narrative around them.
Covalent Metal-Complex Platform Maps Proteome-Wide Cysteines, Nets a PRMT1 Inhibitor
- Introduces reactive metallo-scaffolds (r-mS), a new class of covalent metal complexes bearing a chloroacetamide electrophile designed to engage ligandable cysteines across the proteome.
- Chemoproteomic profiling of a focused r-mS series in HEK293T cell lysate mapped novel cysteine engagement proteome-wide.
- Metal identity, arene substituents, and overall scaffold topography together tune cysteine reactivity and targeting selectivity.
- Lead compound r-mS-2 engages a cysteine within the functionally relevant SAM-binding domain of the methyltransferase PRMT1, confirmed by intact-protein LC-MS.
- That engagement functionally inhibits PRMT1 activity, tying a covalent hit directly to modulation of a validated target.
- Positions the platform as a generalizable route to target discovery for rational, next-generation metallodrug design.
The SAM-pocket cysteine hit on PRMT1 is the part I trust most here — that's a genuinely druggable site, and confirming occupancy by intact-protein LC-MS rather than an activity readout alone is the right bar. Where I'd slow down is selectivity: tuning reactivity through metal identity and arene substituents is a real lever, but it's one more variable than iodoacetamide/chloroacetamide fragment libraries already juggle, and a lysate screen doesn't tell me the off-target cysteine load in cells. Chemically, this is genuinely unfamiliar ground for most of us — ligand exchange, redox lability, and batch-to-batch reproducibility are liabilities organic warheads simply don't have. Against established cysteine-directed chemoproteomic platforms, r-mS adds an interesting third tuning axis, not yet a clear win. Metallodrugs as a class still carry that same tractability tax relative to organic small molecules — worth watching, not yet worth reaching for.
AI-Guided Path to VAV1-Targeting Molecular Glues
- VAV1 is a hematopoietic-restricted signaling protein of interest in blood cancers and autoimmune conditions, prompting efforts to target it via induced protein degradation.
- High-throughput, unbiased global proteomics screening identified phenyl-glutarimide degrader compounds that recruit CRBN to selectively degrade VAV1.
- VAV1's C-terminal SH3-2 domain proved essential for ternary complex formation, and structural work uncovered a novel "RT-loop" degron there—distinct from the "G-loop" degrons used by other CRBN glues—validated by mutagenesis.
- An AI-based protein structure-prediction workflow modeled the ternary complex to support this mechanistic finding.
- Free energy perturbation calculations on the predicted structures generated cooperativity scores that correlated with experimental potency, demonstrating a computational strategy to prioritize and optimize molecular glue analogs faster than empirical screening alone.
The real headline isn't the molecule, it's the RT-loop degron. G-loop CRBN neosubstrates have basically been the only playbook since the IMiD era, so a structurally distinct degron sitting in VAV1's SH3-2 domain is a genuine expansion of the addressable degradable proteome, not just one more target on the pile. The phenyl-glutarimide warhead itself is bog-standard IMiD-adjacent chemistry — nothing novel there; the real innovation is in the target-engagement biology, not the medchem. VAV1's hematopoietic-restricted expression is a nice built-in selectivity advantage most degrader programs would kill for. On the FEP/AI ternary angle: cooperativity scores tracking potency across one series is promising, but it's one dataset — I'd want to see it prospectively rank blind analogs before I trust it over empirical SAR.