Lukas Heumos, Anna C Schaar, Christopher Lance, Anastasia Litinetskaya, Felix Drost, Luke Zappia, Malte D Lücken, Daniel C Strobl, Juan Henao, Fabiola Curion, Single-cell Best Practices Consortium, Herbert B Schiller, Fabian J Theis <<<

Recent advances in single-cell technologies have enabled high-throughput molecular profiling of cells across modalities and locations. Single-cell transcriptomics data can now be complemented by chromatin accessibility, surface protein expression, adaptive immune receptor repertoire profiling and spatial information. The increasing availability of single-cell data across modalities has motivated the development of novel computational methods to help analysts derive biological insights. As the field grows, it becomes increasingly difficult to navigate the vast landscape of tools and analysis steps. Here, we summarize independent benchmarking studies of unimodal and multimodal single-cell analysis across modalities to suggest comprehensive best-practice workflows for the most common analysis steps. Where independent benchmarks are not available, we review and contrast popular methods. Our article serves as an entry point for novices in the field of single-cell (multi-)omic analysis and guides advanced users to the most recent best practices. <<<

Joseph L. Watson , David Juergens , Nathaniel R. Bennett , Brian L. Trippe , Jason Yim , Helen E. Eisenach , Woody Ahern , Andrew J. Borst , Robert J. Ragotte , Lukas F. Milles , Basile I. M. Wicky , Nikita Hanikel , Samuel J. Pellock , Alexis Courbet , William Sheffler , Jue Wang , Preetham Venkatesh , Isaac Sappington , Susana Vázquez Torres , Anna Lauko , Valentin De Bortoli , Emile Mathieu , Regina Barzilay , Tommi S. Jaakkola , Frank DiMaio , Minkyung Baek , David Baker <<<

AbstractThere has been considerable recent progress in designing new proteins using deep learning methods1–9. Despite this progress, a general deep learning framework for protein design that enables solution of a wide range of design challenges, includingde novobinder design and design of higher order symmetric architectures, has yet to be described. Diffusion models10,11have had considerable success in image and language generative modeling but limited success when applied to protein modeling, likely due to the complexity of protein backbone geometry and sequence-structure relationships. Here we show that by fine tuning the RoseTTAFold structure prediction network on protein structure denoising tasks, we obtain a generative model of protein backbones that achieves outstanding performance on unconditional and topology-constrained protein monomer design, protein binder design, symmetric oligomer design, enzyme active site scaffolding, and symmetric motif scaffolding for therapeutic and metal-binding protein design. We demonstrate the power and generality of the method, called RoseTTAFold Diffusion (RFdiffusion), by experimentally characterizing the structures and functions of hundreds of new designs. In a manner analogous to networks which produce images from user-specified inputs, RFdiffusionenables the design of diverse, complex, functional proteins from simple molecular specifications. <<<

The data volume of global information has grown exponentially in recent years, but the development of silicon-based memory has entered a bottleneck period. Deoxyribonucleic acid (DNA) storage is drawing attention owing to its advantages of high storage density, long storage time, and easy maintenance. However, the base utilization and information density of existing DNA storage methods are insufficient. Therefore, this study proposes a rotational coding based on blocking strategy (RBS) for encoding digital information such as text and images in DNA data storage. This strategy satisfies multiple constraints and produces low error rates in synthesis and sequencing. To illustrate the superiority of the proposed strategy, it was compared and analyzed with existing strategies in terms of entropy value change, free energy size, and Hamming distance. The experimental results show that the proposed strategy has higher information storage density and better coding quality in DNA storage, so it will improve the efficiency, practicality, and stability of DNA storage. <<<

Caiying Zhu, Yu Lian, Chenchen Wang, Peng Wu, Xuan Li, Yan Gao, Sibin Fan, Lanlan Ai, Liwei Fang, Hong Pan, Tao Cheng, Jun Shi, Ping Zhu <<<

Aplastic anemia (AA) is a T cell-mediated autoimmune disorder of the hematopoietic system manifested by severe depletion of the hematopoietic stem and progenitor cells (HSPCs). Nonetheless, our understanding of the complex relationship between HSPCs and T cells is still obscure, mainly limited by techniques and the sparsity of HSPCs in the context of bone marrow failure. Here we performed single-cell transcriptome analysis of residual HSPCs and T cells to identify the molecular players from patients with AA. We observed that residual HSPCs in AA exhibited lineage-specific alterations in gene expression and transcriptional regulatory networks, indicating a selective disruption of distinct lineage-committed progenitor pools. In particular, HSPCs displayed frequently altered alternative splicing events and skewed patterns of polyadenylation in transcripts related to DNA damage and repair, suggesting a likely role in AA progression to myelodysplastic syndromes. We further identified cell type-specific ligand-receptor interactions as potential mediators for ongoing HSPCs destruction by T cells. By tracking patients after immunosuppressive therapy (IST), we showed that hematopoiesis remission was incomplete accompanied by IST insensitive interactions between HSPCs and T cells as well as sustained abnormal transcription state. These data collectively constitute the transcriptomic landscape of disrupted hematopoiesis in AA at single-cell resolution, providing new insights into the molecular interactions of engaged T cells with residual HSPCs and render novel therapeutic opportunities for AA. <<<

Yaqi Zhao, Ibrahim Aldoss, Chunxu Qu, Jeremy Chase Crawford, Zhaohui Gu, Emma K Allen, Anthony E Zamora, Thomas B Alexander, Jeremy Wang, Hiroaki Goto, Toshihiko Imamura, Koshi Akahane, Guido Marcucci, Anthony S Stein, Ravi Bhatia, Paul G Thomas, Stephen J Forman, Charles G Mullighan, Kathryn G Roberts <<<

Blinatumomab, a bispecific antibody that directs CD3+ T cells to CD19+ tumor cells, shows variable efficacy in B-progenitor acute lymphoblastic leukemia (B-ALL). To determine tumor-intrinsic and -extrinsic determinants of response, we studied 44 adults with relapsed or refractory B-ALL (including 2 minimal residual disease positive) treated with blinatumomab using bulk tumor and single-cell sequencing. The overall response rate in patients with hematological disease was 55%, with a high response rate in those with CRLF2-rearranged Philadelphia chromosome-like ALL (12 [75%] of 16). Pretreatment samples of responders exhibited a tumor-intrinsic transcriptomic signature of heightened immune response. Multiple mechanisms resulted in loss of CD19 expression, including CD19 mutations, CD19-mutant allele-specific expression, low CD19 RNA expression, and mutations in CD19 signaling complex member CD81. Patients with low hypodiploid ALL were prone to CD19- relapse resulting from aneuploidy-mediated loss of the nonmutated CD19 allele. Increased expression of a CD19 isoform with intraexonic splicing of exon 2, CD19 ex2part, at baseline or during therapy was associated with treatment failure. These analyses demonstrate both tumor-intrinsic and -extrinsic factors influence blinatumomab response. We show that CD19 mutations are commonly detected in CD19- relapse during blinatumomab treatment. Identification of the CD19 ex2part splice variant represents a new biomarker predictive of blinatumomab therapy failure. <<<

Pablo Gainza, Sarah Wehrle, Alexandra Van Hall-Beauvais, Anthony Marchand, Andreas Scheck, Zander Harteveld, Stephen Buckley, Dongchun Ni, Shuguang Tan, Freyr Sverrisson, Casper Goverde, Priscilla Turelli, Charlène Raclot, Alexandra Teslenko, Martin Pacesa, Stéphane Rosset, Sandrine Georgeon, Jane Marsden, Aaron Petruzzella, Kefang Liu, Zepeng Xu, Yan Chai, Pu Han, George F Gao, Elisa Oricchio, Beat Fierz, Didier Trono, Henning Stahlberg, Michael Bronstein, Bruno E Correia <<<

Physical interactions between proteins are essential for most biological processes governing life. However, the molecular determinants of such interactions have been challenging to understand, even as genomic, proteomic and structural data increase. This knowledge gap has been a major obstacle for the comprehensive understanding of cellular protein-protein interaction networks and for the de novo design of protein binders that are crucial for synthetic biology and translational applications. Here we use a geometric deep-learning framework operating on protein surfaces that generates fingerprints to describe geometric and chemical features that are critical to drive protein-protein interactions. We hypothesized that these fingerprints capture the key aspects of molecular recognition that represent a new paradigm in the computational design of novel protein interactions. As a proof of principle, we computationally designed several de novo protein binders to engage four protein targets: SARS-CoV-2 spike, PD-1, PD-L1 and CTLA-4. Several designs were experimentally optimized, whereas others were generated purely in silico, reaching nanomolar affinity with structural and mutational characterization showing highly accurate predictions. Overall, our surface-centric approach captures the physical and chemical determinants of molecular recognition, enabling an approach for the de novo design of protein interactions and, more broadly, of artificial proteins with function. <<<

Seborrheic dermatitis (SD) is the most prevalent dermatological disease, occurring in up to 50% of newborns, children, and adults around the world. The antibacterial and antifungal resistance contributed to the search for new natural substances and the development of a novel substance based on () leaf oil (TTO), 1,8-cineole (eucalyptol), and α-(-)-bisabolol. Thus, this work aimed to determine the chemical composition of the novel plant-based substance and to evaluate its antimicrobial activity against standard microorganisms involved in the pathogenesis of SD. Moreover, the chemical composition of the substance was analyzed by gas chromatography coupled with mass spectrometry (GC/MS). (), (), (), and () were used for antimicrobial and antifungal assays by means of the broth microdilution method to determine the minimal inhibitory concentration (MIC). Finally, the substance's ability to inhibit () was evaluated. Eighteen compounds from different chemical groups were identified by GC/MS. The major biologically active compounds of the substance were terpinen-4-ol (20.88%), 1,8-cineole (22.28%), (-)-α-bisabolol (25.73%), and o-cymene (8.16%). The results showed that the substance has a synergistic antimicrobial and antifungal activity, while and strains were the most susceptible. Furthermore, the substance inhibited , which is a main pathogen involved in the pathogenesis of SD and clinical manifestations. It can be concluded that the novel plant-based substance has a promising potential against and scalp commensal bacteria and may be helpful for the development of new drugs for treatment of dandruff and SD. <<<

For most deep learning practitioners, sequence modeling is synonymous with recurrent networks. Yet recent results indicate that convolutional architectures can outperform recurrent networks on tasks such as audio synthesis and machine translation. Given a new sequence modeling task or dataset, which architecture should one use? We conduct a systematic evaluation of generic convolutional and recurrent architectures for sequence modeling. The models are evaluated across a broad range of standard tasks that are commonly used to benchmark recurrent networks. Our results indicate that a simple convolutional architecture outperforms canonical recurrent networks such as LSTMs across a diverse range of tasks and datasets, while demonstrating longer effective memory. We conclude that the common association between sequence modeling and recurrent networks should be reconsidered, and convolutional networks should be regarded as a natural starting point for sequence modeling tasks. To assist related work, we have made code available at this http URL . <<<

Jianming Yang, Geng Pei, Xuan Sun, Yawen Xiao, Chunhui Miao, Lu Zhou, Bangmao Wang, Liu Yang, Mingyu Yu, Zhi-Song Zhang, Evan T Keller, Zhi Yao, Quan Wang <<<

BACKGROUND: The pathogenesis of inflammatory bowel diseases (IBD) is multifactorial, and diagnostic and treatment strategies for IBD remain to be developed. RhoB regulates multiple cell functions; however, its role in colitis is unexplored.RESULTS: Here, we found RhoB was dramatically increased in colon tissues of ulcerative colitis (UC) patients and mice with DSS-induced colitis. Compared with wild type mice, RhoB+/- and RhoB-/- mice developed milder DSS-induced colitis and increased goblet cell numbers and IEC proliferation. Decreased RhoB promoted goblet cell differentiation and epithelial regeneration through inhibiting Wnt signaling pathway and activating p38 MAPK signaling pathway. Moreover, increased SCFA-producing bacteria and SCFA concentrations were detected in intestinal microbiome of both RhoB+/- and RhoB-/- mice and upregulated SCFA receptor expression was also observed.CONCLUSIONS: Taken together, a higher level of RhoB is associated with UC, which also contributes to UC development through modulating cell signaling and altering intestinal bacterial composition and metabolites. These observations suggest that RhoB has potential as a biomarker and a treatment target for UC. Video Abstract. <<<

In response to physical exercise and diet, skeletal muscle adapts to energetic demands through large transcriptional changes. This remodelling is associated with changes in skeletal muscle DNA methylation which may participate in the metabolic adaptation to extracellular stimuli. Yet, the mechanisms by which muscle-borne DNA methylation machinery responds to diet and exercise and impacts muscle function are unknown. Here, we investigated the function of de novo DNA methylation in fully differentiated skeletal muscle. We generated muscle-specific DNA methyltransferase 3A (DNMT3A) knockout mice (mD3AKO) and investigated the impact of DNMT3A ablation on skeletal muscle DNA methylation, exercise capacity and energy metabolism. Loss of DNMT3A reduced DNA methylation in skeletal muscle over multiple genomic contexts and altered the transcription of genes known to be influenced by DNA methylation, but did not affect exercise capacity and whole-body energy metabolism compared to wild type mice. Loss of DNMT3A did not alter skeletal muscle mitochondrial function or the transcriptional response to exercise however did influence the expression of genes involved in muscle development. These data suggest that DNMT3A does not have a large role in the function of mature skeletal muscle although a role in muscle development and differentiation is likely. <<<

Ali Madani, Ben Krause, Eric R Greene, Subu Subramanian, Benjamin P Mohr, James M Holton, Jose Luis Olmos, Caiming Xiong, Zachary Z Sun, Richard Socher, James S Fraser, Nikhil Naik <<<

Deep-learning language models have shown promise in various biotechnological applications, including protein design and engineering. Here we describe ProGen, a language model that can generate protein sequences with a predictable function across large protein families, akin to generating grammatically and semantically correct natural language sentences on diverse topics. The model was trained on 280 million protein sequences from >19,000 families and is augmented with control tags specifying protein properties. ProGen can be further fine-tuned to curated sequences and tags to improve controllable generation performance of proteins from families with sufficient homologous samples. Artificial proteins fine-tuned to five distinct lysozyme families showed similar catalytic efficiencies as natural lysozymes, with sequence identity to natural proteins as low as 31.4%. ProGen is readily adapted to diverse protein families, as we demonstrate with chorismate mutase and malate dehydrogenase. <<<

State-of-the-art computer vision systems are trained to predict a fixed set of predetermined object categories. This restricted form of supervision limits their generality and usability since additional labeled data is needed to specify any other visual concept. Learning directly from raw text about images is a promising alternative which leverages a much broader source of supervision. We demonstrate that the simple pre-training task of predicting which caption goes with which image is an efficient and scalable way to learn SOTA image representations from scratch on a dataset of 400 million (image, text) pairs collected from the internet. After pre-training, natural language is used to reference learned visual concepts (or describe new ones) enabling zero-shot transfer of the model to downstream tasks. We study the performance of this approach by benchmarking on over 30 different existing computer vision datasets, spanning tasks such as OCR, action recognition in videos, geo-localization, and many types of fine-grained object classification. The model transfers non-trivially to most tasks and is often competitive with a fully supervised baseline without the need for any dataset specific training. For instance, we match the accuracy of the original ResNet-50 on ImageNet zero-shot without needing to use any of the 1.28 million training examples it was trained on. We release our code and pre-trained model weights at this https URL. <<<

An established normative approach for understanding the algorithmic basis of neural computation is to derive online algorithms from principled computational objectives and evaluate their compatibility with anatomical and physiological observations. Similarity matching objectives have served as successful starting points for deriving online algorithms that map onto neural networks (NNs) with point neurons and Hebbian/anti-Hebbian plasticity. These NN models account for many anatomical and physiological observations; however, the objectives have limited computational power and the derived NNs do not explain multi-compartmental neuronal structures and non-Hebbian forms of plasticity that are prevalent throughout the brain. In this article, we review and unify recent extensions of the similarity matching approach to address more complex objectives, including a broad range of unsupervised and self-supervised learning tasks that can be formulated as generalized eigenvalue problems or nonnegative matrix factorization problems. Interestingly, the online algorithms derived from these objectives naturally map onto NNs with multi-compartmental neurons and local, non-Hebbian learning rules. Therefore, this unified extension of the similarity matching approach provides a normative framework that facilitates understanding the multi-compartmental neuronal structures and non-Hebbian plasticity found throughout the brain. <<<

INTRODUCTION: The impact of pectoralis muscle mass index (PMI) on cardiac events is not well studied in cancer patients, especially in those who have received chemotherapy with high potential cardiac toxicity such as anthracyclines.METHODS: Individuals aged ≥18 years with a diagnosis of breast cancer, sarcoma, or lymphoma who received anthracycline-based chemotherapy at the University of Minnesota MHealth Fairview between 2009 and 2014. Eligible patients had to have two CT scans: a baseline CT scan within 6 months prior to chemotherapy and a follow-up CT scan within 2 years after treatment. The PMI was calculated as the right pectoralis muscle area indexed to height squared. Multivariable linear regression was used to analyze factors associated with PMI at follow-up, overall mortality, and major cardiac events (MACE).RESULTS: A total of 474 patients (breast cancer 192; lymphoma 184; sarcoma 98) participated with a median age of 61 years at the time of baseline CT scan; 161 (34%) were male. Almost all patients received anthracyclines except 12% who received trastuzumab only. The median baseline PMI was 5.8 cm2/m2 (4.9, 7.7) which decreased 10.5% after chemotherapy, to 5.2 cm2/m2 (4.4, 6.4). Baseline PMI was not significantly associated with OS, but we detected lower risks of MACE with larger PMI at baseline. Greater baseline PMI was associated with greater follow-up PMI, but also with greater relative PMI loss. Female gender, older age, and history of smoking were also associated with greater PMI losses.CONCLUSION: Greater pre-treatment pectoralis muscle index in patients treated with anthracyclines have a lower risk of MACE. Early identification of sarcopenia using PMI could trigger proactive engagement for intervention and risk-stratified therapies. <<<

Perceptual events derive their significance to an animal from their meaning about the world, that is from the information they carry about their causes. The brain should thus be able to efficiently infer the causes underlying our sensory events. Here we use multisensory cue combination to study causal inference in perception. We formulate an ideal-observer model that infers whether two sensory cues originate from the same location and that also estimates their location(s). This model accurately predicts the nonlinear integration of cues by human subjects in two auditory-visual localization tasks. The results show that indeed humans can efficiently infer the causal structure as well as the location of causes. By combining insights from the study of causal inference with the ideal-observer approach to sensory cue combination, we show that the capacity to infer causal structure is not limited to conscious, high-level cognition; it is also performed continually and effortlessly in perception. <<<

The role of enhancers, a key class of non-coding regulatory DNA elements, in cancer development has increasingly been appreciated. Here, we present the detection and characterization of a large number of expressed enhancers in a genome-wide analysis of 8928 tumor samples across 33 cancer types using TCGA RNA-seq data. Compared with matched normal tissues, global enhancer activation was observed in most cancers. Across cancer types, global enhancer activity was positively associated with aneuploidy, but not mutation load, suggesting a hypothesis centered on "chromatin-state" to explain their interplay. Integrating eQTL, mRNA co-expression, and Hi-C data analysis, we developed a computational method to infer causal enhancer-gene interactions, revealing enhancers of clinically actionable genes. Having identified an enhancer ∼140 kb downstream of PD-L1, a major immunotherapy target, we validated it experimentally. This study provides a systematic view of enhancer activity in diverse tumor contexts and suggests the clinical implications of enhancers. <<<

The effects of pasteurisation, high-pressure (HP), and a combination of pasteurisation and high-pressure (PHP) on the physicochemical properties and protein structure of caprine skim milk was investigated. Samples treated by PHP generally had a higher pH, whey protein denaturation, surface hydrophobicity, and intrinsic fluorescence than those treated only with heat or pressure. In contrast, the size of skim milk casein micelles decreased significantly with an increase in pressure level and time; however, the effect was less marked when heat and pressure treatments were combined. For the PHP and HP samples, as the level and time of pressure increased, the α-helix and β-turn content reduced, whereas β-sheet and random coil were induced. Thus, PHP treatment could be used as a good alternative technology in the dairy industry to promote the functional properties of skim caprine milk. <<<

The use of NLP in the realm of financial technology is broad and complex, with applications ranging from sentiment analysis and named entity recognition to question answering. Large Language Models (LLMs) have been shown to be effective on a variety of tasks; however, no LLM specialized for the financial domain has been reported in literature. In this work, we present BloombergGPT, a 50 billion parameter language model that is trained on a wide range of financial data. We construct a 363 billion token dataset based on Bloomberg's extensive data sources, perhaps the largest domain-specific dataset yet, augmented with 345 billion tokens from general purpose datasets. We validate BloombergGPT on standard LLM benchmarks, open financial benchmarks, and a suite of internal benchmarks that most accurately reflect our intended usage. Our mixed dataset training leads to a model that outperforms existing models on financial tasks by significant margins without sacrificing performance on general LLM benchmarks. Additionally, we explain our modeling choices, training process, and evaluation methodology. As a next step, we plan to release training logs (Chronicles) detailing our experience in training BloombergGPT. <<<

OBJECTIVES: Bacteriome and virome alterations are associated with colorectal cancer (CRC). Nevertheless, the gut fungal microbiota in CRC remains largely unexplored. We aimed to characterise enteric mycobiome in CRC.DESIGN: Faecal shotgun metagenomic sequences of 184 patients with CRC, 197 patients with adenoma and 204 control subjects from Hong Kong were analysed (discovery cohort: 73 patients with CRC and 92 control subjects; validation cohort: 111 patients with CRC, 197 patients with adenoma and 112 controls from Hong Kong). CRC-associated fungal markers and ecological changes were also validated in additional independent cohorts of 90 patients with CRC, 42 patients with adenoma and 66 control subjects of published repository sequences from Germany and France. Assignment of taxonomies was performed by exact k-mer alignment against an integrated microbial reference genome database.RESULTS: Principal component analysis revealed separate clusters for CRC and control (p<0.0001), with distinct mycobiomes in early-stage and late-stage CRC (p=0.0048). Basidiomycota:Ascomycota ratio was higher in CRC (p=0.0042), with increase in Malasseziomycetes (p<0.0001) and decrease in Saccharomycetes (p<0.0001) and Pneumocystidomycetes (p=0.0017). Abundances of 14 fungal biomarkers distinguished CRC from controls with an area under the receiver-operating characteristic curve (AUC) of 0.93 and validated AUCs of 0.82 and 0.74 in independent Chinese cohort V1 and European cohort V2, respectively. Further ecological analysis revealed higher numbers of co-occurring fungal intrakingdom and co-exclusive bacterial-fungal correlations in CRC (p<0.0001). Moreover, co-occurrence interactions between fungi and bacteria, mostly contributed by fungal Ascomycota and bacterial Proteobacteria in control, were reverted to co-exclusive interplay in CRC (p=0.00045).CONCLUSIONS: This study revealed CRC-associated mycobiome dysbiosis characterised by altered fungal composition and ecology, signifying that the gut mycobiome might play a role in CRC. <<<