Relapse and refractory T-cell acute lymphoblastic leukemia (T-ALL) has a poor prognosis, and new combination therapies are sorely needed. Here, we used an ex vivo high-throughput screening platform to identify drug combinations that kill zebrafish T-ALL and then validated top drug combinations for preclinical efficacy in human disease. This work uncovered potent drug synergies between AKT/mTORC1 (mammalian target of rapamycin complex 1) inhibitors and the general tyrosine kinase inhibitor dasatinib. Importantly, these same drug combinations effectively killed a subset of relapse and dexamethasone-resistant zebrafish T-ALL. Clinical trials are currently underway using the combination of mTORC1 inhibitor temsirolimus and dasatinib in other pediatric cancer indications, leading us to prioritize this therapy for preclinical testing. This combination effectively curbed T-ALL growth in human cell lines and primary human T-ALL and was well tolerated and effective in suppressing leukemia growth in patient-derived xenografts (PDX) grown in mice. Mechanistically, dasatinib inhibited phosphorylation and activation of the lymphocyte-specific protein tyrosine kinase (LCK) to blunt the T-cell receptor (TCR) signaling pathway, and when complexed with mTORC1 inhibition, induced potent T-ALL cell killing through reducing MCL-1 protein expression. In total, our work uncovered unexpected roles for the LCK kinase and its regulation of downstream TCR signaling in suppressing apoptosis and driving continued leukemia growth. Analysis of a wide array of primary human T-ALLs and PDXs grown in mice suggest that combination of temsirolimus and dasatinib treatment will be efficacious for a large fraction of human T-ALLs.

Linked-read sequencing was developed to aid the detection of large structural variants (SVs) from short-read sequencing efforts. We performed a systematic evaluation to determine if linked-read exome sequencing provides more comprehensive and clinically relevant information than whole-exome sequencing (WES) when applied to the same set of multiple myeloma patient samples. We report that linked-read sequencing detected a higher number of SVs (n = 18,455) than WES (n = 4065). However, linked-read predictions were dominated by inversions (92.4%), leading to poor detection of other types of SVs. In contrast, WES detected 56.3% deletions, 32.6% insertions, 6.7% translocations, 3.3% duplications and 1.2% inversions. Surprisingly, the quantitative performance assessment suggested a higher performance for WES (AUC = 0.791) compared to linked-read sequencing (AUC = 0.766) for detecting clinically validated cytogenetic alterations. We also found that linked-read sequencing detected more short variants (n = 704) compared to WES (n = 109). WES detected somatic mutations in all MM-related genes while linked-read sequencing failed to detect certain mutations. The comparison of somatic mutations detected using linked-read, WES and RNA-seq revealed that WES and RNA-seq detected more mutations than linked-read sequencing. These data indicate that WES outperforms and is more efficient than linked-read sequencing for detecting clinically relevant SVs and MM-specific short variants.

Background Drug-tolerant persister (DTP) cell states have been implicated in relapse across multiple cancers, including acute myeloid leukaemia (AML) [1,2]. Methods that score such states from transcriptomic data, generalise to held-out samples, expose calibrated probability outputs, and link predictions to candidate biology are useful for prioritising follow-up experimental work. Existing transcriptomic methods for scoring drug-tolerant or persister-like states largely rely on fixed gene signatures or general-purpose cell-type classifiers adapted post hoc (scPred, scANVI, scClassify); deep-learning approaches developed specifically for AML drug-tolerant persister scoring with calibrated probability outputs, prespecified thresholds, and transparent external validation against ex-vivo drug-response data are, to our knowledge, lacking. Our approach addresses this gap by combining a Transformer teacher with a knowledge-distilled 1,000-gene student, prespecified threshold τ = 0.31, and direct evaluation against BeatAML drug-AUC. Our in silico approach aims to fill this gap of non-existent analytical methods to identify and mark the DTP cells. Methods We trained a Transformer classifier on a pooled scRNA-seq corpus of nine samples (six from GSE123902 -lung adenocarcinoma metastasis, normal, and primary tumour [4] -plus three primary AML samples; 32,342 cells, 13,369 common genes), with stratified 5-fold cross-validation at the cell level, a 20% held-out test split, and a prespecified probability threshold selected on out-of-fold predictions. A 1,000-gene student model was trained by knowledge distillation [5]. For every input cell, the student outputs a probability between 0 and 1 (hereafter "the score") representing predicted membership in the positive training class. The trained model was applied without re-tuning to five external or independent application cohorts: 39 primary AML donors[in-house]; GSE74246[6]; BeatAML (n = 452 with linked ex-vivo drug-AUC; n = 405 with overall-survival metadata)[7]; TCGA-LAML (n = 149)[8]; and an in-house n = 10 scRNA-seq cohort with linked survival. Survival and drug-response data were not used during training, threshold selection, or tuning. The score was anchored mechanistically against CRISPR/DepMap essentiality[9], pathway enrichment, and a normal-tissue-filtered surface-protein candidate list (HPA[11], GTEx[12]). To assess concordance between transcriptomic prioritisation and protein-level evidence, each ranked candidate was additionally annotated with two HPA-derived flags: HPA_surface_protein (Yes/No, derived from HPA Protein class and Subcellular location fields, identifying genes annotated as plasma-membrane, GPCR, ion-channel, transporter, receptor, or CD-marker) and HPA_antibody_reliability (Enhanced, Supported, Approved, Uncertain, or Not available, per HPA antibody validation tier). Annotations were merged on HGNC symbol; 248 of 250 candidates (99.2%) matched. Two candidates using the older CORF nomenclature did not auto-match HPA's lowercase convention and were resolved manually. HPA's per-gene RNA-protein numeric correlation is published only on per-gene web pages and not in the bulk download; we therefore used the detection-level and antibody-reliability tiers as the operational concordance filter. Results Cross-validation area under the receiver operating characteristic curve (AUROC) was 0.936 +/- 0.014 (held-out test 0.941, Matthews correlation coefficient (MCC) 0.696, F1-score 0.895). The 1,000-gene student showed Spearman ρ ≈ 0.96 with the teacher and >85% class agreement at the prespecified threshold. The principal external result was in BeatAML: the score correlated with ex-vivo drug-response AUC across seven AML-relevant drugs, with consistent per-drug Spearman correlations (r = 0.41-0.53, all p < 0.05). The aggregate correlation across 3,164 patient-drug pairs from 452 patients was r = +0.482 and is reported as a summary, recognising that pairs from the same patient are not fully independent. The score did not stratify overall survival in TCGA-LAML or in the in-house n = 10 cohort, in part because predicted high-score fractions saturated. At the prespecified threshold the score did not separate cell types in GSE74246, indicating that absolute calibration is cohort-dependent. Compared against logistic regression, random forest, the LSC17 stemness signature, and a mean-expression baseline on the same gene panel, the Transformer was the most stable model under aliquot-grouped cross-validation and the only one to transfer with strong, positive correlation to BeatAML drug-AUC. The mechanistic candidate-target pipeline produced a 250-candidate ranked surface-protein list (full breakdown in Results); FLT3 and CD33 were recovered from the unbiased ranking as positive controls. Conclusion We present a Transformer-derived transcriptomic score that addresses the lack of validated computational methods for identifying drug-tolerant persister-like states in AML. The score shows external rank-order association with ex-vivo drug response, providing a research-use tool for prioritising candidate persister-associated transcriptional programs for follow-up. Together, these results support the score as a research-use transcriptomic ranking tool for AML drug-response-associated states. The strongest external support comes from the consistent association with BeatAML ex-vivo drug-response AUC. The fixed probability threshold did not transfer reliably across all cohorts, so threshold-based classification should require cohort-specific recalibration. The score is not validated for clinical decision-making and is not proposed as a survival predictor. The candidate-target list is a starting point for functional follow-up. Keywords. AML; ex-vivo drug response; single-cell RNA-seq; Transformer; knowledge distillation; transcriptomic score; BeatAML; surface-protein target prioritisation.

Background: Acute myeloid leukemia (AML) progresses by the accumulation of somatic mutations and clonal expansion of pre-leukemic cells. Patients may respond to initial therapy, but often relapse, underscoring an evolving disease. Next generation sequencing technologies are being applied to AML for risk stratification and monitoring treatment response. We aimed to evaluate the efficiency of whole exome sequencing (WES) and single-cell DNA sequencing (scDNA-seq) for determining clonal heterogeneity and evolution in AML induced by treatment, assessing strengths and limitations of each technology. Methods: We conducted WES and scDNA-seq on samples from 6 patients with AML, including sequential samples from four patients. We identified somatic variants, clonal composition and phylogeny using both technologies and compared the results. Results: WES detected more variants and clones due to broader coverage, while scDNA-seq provided clonality results for targeted genes revealing zygosity and rare clones. Both techniques missed clinically important variants, posing challenges for clinical application. However, they identified similar founding clones and strong correlation of variant allele frequencies and clonal prevalences. Conclusions: As both technologies can overlook variants, multiple technologies should be utilized to understand clonality in heterogeneous diseases such as AML. Careful scDNA-seq target panel planning, utilizing knowledge obtained from bulk sequencing, can offer more information on clonal heterogeneity. Graphical

ERCC6L2 disease (ED) is a rare bone marrow failure syndrome caused by biallelic germline mutations in ERCC6L2 . ED leads to the accumulation of somatic TP53 mutations, myelodysplastic syndrome, and acute myeloid leukemia (AML) with erythroid predominance and poor prognosis. While ERCC6L2 is implicated in DNA replication and repair, the transcriptomic events underlying delayed erythropoiesis and leukemic progression remain largely undefined. To delineate these processes, we leverage bulk and single‐cell transcriptomics of patient fibroblasts, bone marrow, and peripheral blood across disease stages, including single‐cell TP53 genotyping. We identify disease‐associated erythroid dysregulation and ferroptotic stress emerging prior to TP53 mutation, highlighting an early vulnerability in ED leukemogenesis. We compare ED to Shwachman–Diamond syndrome (SDS) to reveal shared and disease‐specific transcriptional programs. TP53 mutations in ED and SDS arise in hematopoietic stem and progenitor cells but do not independently drive changes in cell cycle or stress pathways during erythropoiesis, despite harboring distinct germline defects. Both diseases converge in late erythropoiesis into a stress state characterized by ferroptotic signaling, G1 arrest, and BCL2L1 upregulation. As a disease‐specific pattern, ED shows aberrant erythroid priming with TP53 ‐driven differentiation arrest shaping progression toward erythroid leukemia. Thereby, we establish the first patient‐level single‐cell map of ED and provide a curated resource for future work on ED, SDS, and TP53 ‐driven leukemogenesis. Overall, our findings in pre‐malignant ED offer a window into early alterations leading to high‐risk leukemia.