Biology subjects

Feng, F.

Publications and source records attributed to Feng, F..

At least 37 records · Page 2Linked to original sources

bioRxiv2024

Better Prior Distribution for Antibody Complementarity-Determining Regions Design

Designing antibodies with desired binding specificity and affinity is essential for pharmaceutical research. While diffusion-based models have advanced the co-design of the Complementarity-Determining Regions (CDRs) sequences and structures, challenges remain, including non-informative prior distribution, incompatibility with discrete amino acid types, and impractical computational cost in large-scale sampling. To address these, we proposed FlowDesign, a sequence-structure co-design approach based on Flow Matching, offering: (1) Flexible selection of prior distributions; (2) Direct matching of discrete distributions; (3) Enhanced computational efficiency for large-scale sampling. By leveraging various priors, data-driven structural models proved the most informative. FlowDesign outperformed baselines in Amino Acid Recovery (AAR), RMSD, and Rosetta energy. We also applied FlowDesign to design antibodies targeting the HIV-1 receptor CD4. FlowDesign yielded antibodies with improved binding affinity and neutralizing potency compared to the antibody Ibalizumab across multiple HIV mutants, validated by Biolayer Interferometry (BLI) and pseudovirus neutralization. This highlights FlowDesigns potential in antibody and protein design. A record of this papers Transparent Peer Review process is included in the Supplemental Information.

bioinformatics↗

bioRxiv2024

The Medicago truncatula LYR4 intracellular domain serves as a scaffold in immunity signaling independent of its phosphorylation activity

Plants perceive and respond to chitin derived from fungal cell walls through lysine motif (LysM) receptor kinases. In the model legume Medicago truncatula, CERK1 and LYR4 represent the LysM receptor pair important for chitin-triggered immunity signaling. Here, we show that both the active kinase receptor CERK1 and the pseudokinase receptor LYR4 contribute to immunity signaling, leading to the production of reactive oxygen species (ROS). We determine the crystal structure of the LYR4 core intracellular domain with a bound nucleotide analog in the active site. Biochemical characterization shows that LYR4 binds ATP and has both autophosphorylation as well as transphosphorylation activity towards CERK1. However, in planta experiments demonstrate that the phosphorylation ability is not necessary for the function of LYR4 in chitin-triggered ROS production, but that the presence of its intracellular domain is indispensable. Together, we show that in chitin-triggered immunity the intracellular domain of LYR4 serves as a signaling scaffold independent of its catalytic activity.

plant biology↗

bioRxiv2024

Building a literature knowledge base towards transparent biomedical AI

As artificial intelligence (AI) continues to advance and scale up in biomedical research, concerns about AIs trustworthiness and transparency have grown. There is a critical need to systematically bring accurate and relevant biomedical knowledge into AI applications for transparency and provenance. Knowledge graphs have emerged as a powerful tool that integrates heterogeneous knowledge by explicitly describing biomedical knowledge as entities and relationships between entities. However, PubMed, the largest and most comprehensive repository of biomedical knowledge, exists primarily as unstructured text and is under utilized for advanced machine learning tasks. To address the challenge, we developed LiteralGraph, a computational framework to extract biomedical terms and relationships from PubMed literature into a unified knowledge graph. Using this framework, we established the Genomic Literature Knowledge Base (GLKB), which consolidates 14,634,427 biomedical relationships between 3,276,336 biomedical terms from over 33 million PubMed abstracts and nine well-established biomedical repositories. The database is coupled with RESTful APIs and a user-friendly web interface that makes it accessible to researchers for various usages. We demonstrated the broad utility of GLKB towards transparent AI in three distinct application scenarios. In the LLM grounding scenario, we developed a Retrieval Augmented Generation (RAG) agent to reduce LLM hallucination in biomedical question answering. In the hypothesis generation scenario, we elucidated the potential functions of RFX6 in type 2 diabetes (T2D) using the vast evidence from PubMed articles. In the machine learning scenario, we utilized GLKB to provide semantic knowledge in predictive tasks and scientific fact-checking.

bioinformatics↗

bioRxiv2024

Dual-view jointly learning improves personalized drug synergy prediction

BackgroundAccurate and robust estimation of the synergistic drug combination is important for precision medicine. Although some computational methods have been developed, some predictions are still unreliable especially for the cross-dataset predictions, due to the complex mechanism of drug combinations and heterogeneity of cancer samples. MethodsWe have proposed JointSyn that utilizes dual-view jointly learning to predict sample-specific effects of drug combination from drug and cell features. JointSyn capture the drug synergy related features from two views. One view is the embedding of drug combination on cancer cell lines, and the other view is the combination of two drugs embeddings on cancer cell lines. Finally, the prediction net uses the features learned from the two views to predict the drug synergy of the drug combination on the cell line. In addition, we used the fine-tuning method to improve the JointSyns performance on the unseen subset within a dataset or cross dataset. ResultsJointSyn outperforms existing state-of-the-art methods in predictive accuracy and robustness across various benchmarks. Each view of JointSyn captures drug synergy-related characteristics and make complementary contributes to the final accurate prediction of drug combination. Moreover, JointSyn with fine-tuning improves its generalization ability to predict a novel drug combination or cancer sample only using a small number of experimental measurements. We also used JointSyn to generate an estimated atlas of drug synergy for pan-cancer and explored the differential pattern among cancers. ConclusionsThese results demonstrate the potential of JointSyn to predict drug synergy, supporting the development of personalized combinatorial therapies. The source code is available on GitHub at https://github.com/LiHongCSBLab/JointSyn.

cancer biology↗

bioRxiv2023

Population-scale skeletal muscle single-nucleus multi-omic profiling reveals extensive context specific genetic regulation

Skeletal muscle, the largest human organ by weight, is relevant in several polygenic metabolic traits and diseases including type 2 diabetes (T2D). Identifying genetic mechanisms underlying these traits requires pinpointing cell types, regulatory elements, target genes, and causal variants. Here, we use genetic multiplexing to generate population-scale single nucleus (sn) chromatin accessibility (snATAC-seq) and transcriptome (snRNA-seq) maps across 287 frozen human skeletal muscle biopsies representing nearly half a million nuclei. We identify 13 cell types and integrate genetic variation to discover >7,000 expression quantitative trait loci (eQTL) and >100,000 chromatin accessibility QTLs (caQTL) across cell types. Learning patterns of e/caQTL sharing across cell types increased precision of effect estimates. We identify high-resolution cell-states and context-specific e/caQTL with significant genotype by context interaction. We identify nearly 2,000 eGenes colocalized with caQTL and construct causal directional maps for chromatin accessibility and gene expression. Almost 3,500 genome-wide association study (GWAS) signals across 38 relevant traits colocalize with sn-e/caQTL, most in a cell-specific manner. These signals typically colocalize with caQTL and not eQTL, highlighting the importance of population-scale chromatin profiling for GWAS functional studies. Finally, our GWAS-caQTL colocalization data reveal distinct cell-specific regulatory paradigms. Our results illuminate the genetic regulatory architecture of human skeletal muscle at high resolution epigenomic, transcriptomic, and cell-state scales and serve as a template for population-scale multi-omic mapping in complex tissues and traits.

genomics↗

bioRxiv2023

ZNF397 Loss Triggers TET2-driven Epigenetic Rewiring, Lineage Plasticity, and AR-targeted Therapy Resistance in AR-dependent Cancers

Cancer cells exhibit phenotypical plasticity and epigenetic reprogramming, which allows them to evade lineage-dependent targeted treatments by adopting lineage plasticity. The underlying mechanisms by which cancer cells exploit the epigenetic regulatory machinery to acquire lineage plasticity and therapy resistance remain poorly understood. We identified Zinc Finger Protein 397 (ZNF397) as a bona fide co-activator of the androgen receptor (AR), essential for the transcriptional program governing AR-driven luminal lineage. ZNF397 deficiency facilitates the transition of cancer cell from an AR-driven luminal lineage to a Ten-Eleven Translocation 2 (TET2)-driven lineage plastic state, ultimately promoting resistance to therapies inhibiting AR signaling. Intriguingly, our findings indicate that TET2 inhibitor can eliminate the AR targeted therapies resistance in ZNF397-deficient tumors. These insights uncover a novel mechanism through which prostate and breast cancers acquire lineage plasticity via epigenetic rewiring and offer promising implications for clinical interventions designed to overcome therapy resistance dictated by lineage plasticity. Statement of SignificanceThis study reveals a novel epigenetic mechanism regulating tumor lineage plasticity and therapy response, enhances understanding of drug resistance and unveils a new therapeutic strategy for prostate cancer and other malignancies. Our findings also illuminate TET2s oncogenic role and mechanistically connect TET2-driven epigenetic rewiring to lineage plasticity and therapy resistance.

cancer biology↗

bioRxiv2023

Marmoset Brain Mapping V5: an Anatomical and Connectivity Atlas of the Cerebellum

The cerebellum is a key region of the brain for motor control and cognitive functioning, and it is involved bidirectional communication with the cerebral cortex. As one of the smallest non-human primates, the common marmoset provides many advantages in the study of the anatomy and the functions of cerebello-cerebral circuits. However, the cerebellum of the marmoset is far from being well described in published resources. In this study, we present a comprehensive atlas of the marmoset cerebellum in three parts: 1) fine-detailed anatomical atlases and surface-analyzing tools of the cerebellar cortex, based on ultra-high resolution ex-vivo MRI; 2) functional-connectivity gradients of the cerebellar cortex, based on awake resting-state fMRI; and 3) structural-connectivity based mapping of the cerebellar nuclei, based on high-resolution diffusion tractography. The atlas shows the anatomical details of the marmoset cerebellum; reveals distinct gradient patterns of the intra-cerebellar and the cerebello-cerebral functional connectivity; and maps the topological relationship of the cerebellar nuclei in cerebello-cerebral circuits. As version 5 of the Marmoset Brain Mapping project, the atlas describes the anatomical and connectivity details of the marmoset cerebellum and is publicly available via marmosetbrainmapping.org.

neuroscience↗

bioRxiv2022

Characterizing adjuvants' effects at the murine immunoglobulin repertoire level

High-throughput immunoglobulin sequencing (IgSeq) has been developed and applied to study the adaptive immune response extensively for more than a decade. However, generating large-scale, high-fidelity sequencing data is still challenging, and furthermore, not much has been done to characterize adjuvants effects at the repertoire level. Thus, we developed an improved library prep protocol and standardized the data analysis pipeline for accurate repertoire profiling. In addition, two metrics were implemented to assess repertoire clone properties. We then studied systemically the effects of two adjuvants, CpG and Alum, on the Ig heavy chain repertoire using the ovalbumin (OVA) challenged mouse model. Ig repertoires of different tissues (spleen and bone marrow) and isotypes (IgG and IgM) were examined and compared in terms of sequence mutation frequency, IGHV gene usage, CDR3 length, rescaled Hill numbers for clonal diversity, and clone selection strength. As a result, Ig repertoires of different tissues or isotypes exhibited distinguishable profiles at the non-immunized steady state. Adjuvanted immunizations further resulted in statistically significant alterations in Ig repertoire compared with PBS or OVA alone immunized groups. Lastly, we applied unsupervised machine learning techniques - multiple factor analysis and clustering - to identify Ig repertoire signatures in different compartments and under varying immunizations. We found that the IGH repertoires of distinct tissue-isotype compartments or under varying immunizations differed in unique sets of properties. Notably, Alum and CpG effects on the Ig repertoire exhibited different tissue and isotype preferences. The former led to increased diversity of abundant clones of both isotypes in BM only, and the latter promoted the selection of IgG clones only but in both tissues. The patterns of Ig repertoire changes likely reflected possible action mechanisms of these two adjuvants.

immunology↗

bioRxiv2022

Noninvasive cancer detection by extracting and integrating multi-modal data from whole-methylome sequencing of plasma cell-free DNA

AbstractPlasma cell-free DNA (cfDNA) methylation and fragmentation signatures have been shown to be valid biomarkers for blood-based cancer detection. However, conventional methylation sequencing assays are inapplicable for fragmentomic profiling due to bisulfite-induced DNA damage. Here using enzymatic conversion-based low-pass whole-methylome sequencing (WMS), we developed a novel approach to comprehensively interrogate the genome-wide plasma methylation, fragmentation, and copy number profiles for sensitive and noninvasive multi-cancer detection. With plasma WMS data from a clinical cohort comprising 497 healthy controls and 780 patients with both early- and advanced-stage cancers of the breast, colorectum, esophagus, stomach, liver, lung, or pancreas, genomic features including methylation, fragmentation size, copy number alteration, and fragment end motif were extracted individually and subsequently integrated to develop an ensemble cancer classifier, called THEMIS, using machine learning algorithms. THEMIS outperformed individual biomarkers for differentiating cancer patients of all seven types from healthy individuals and achieved a combined area under the curve value of 0.971 in the independent test cohort, translating to a sensitivity of 86% and early-stage (I and II) sensitivity of 77% at 99% specificity. In addition, we built a cancer signal origin classifier with true-positive cancer samples at 100% specificity based on methylation and fragmentation profiling of tissue-specific accessible regulatory elements, which localized cancer-like signal to a limited number of clinically informative sites with 66% accuracy. Overall, this proof-of-concept work demonstrates the feasibility of extracting and integrating multi-modal biomarkers from a single WMS run for noninvasive detection and localization of common cancers across stages.

cancer biology↗

bioRxiv2022

A generalizable framework to comprehensively predict epigenome, chromatin organization, and transcriptome

Many deep learning approaches have been proposed to predict epigenetic profiles, chromatin organization, and transcription activity. While these approaches achieve satisfactory performance in predicting one modality from another, the learned representations are not generalizable across predictive tasks or across cell types. In this paper, we propose a deep learning approach named EPCOT which employs a pre-training and fine-tuning framework, and comprehensively predicts epigenome, chromatin organization, transcriptome, and enhancer activity in one framework. EPCOT is the first framework proposed to predict all of these genomic modalities and performs well in individual modality prediction, which is also generalizable to new cell and tissue types. EPCOT also maps from DNA sequence and chromatin accessibility profiles to generic representations which are generalizable across different modalities. Interpreting EPCOT model also provides biological insights including mapping between different genomic modalities, identifying TF sequence binding patterns, and analyzing cell-type specific TF impacts on enhancer activity.

bioinformatics↗

bioRxiv2022

Identification and calibration of ultrabright localizations to eliminate quantification error in SMLM

Single molecule localization microscopy (SMLM) is irreplaceable among super-resolution microscopies in revealing biological ultra-structures, given its unmatched high resolution. However, its sub-optimal quantitative capability, which is critical for characterizing true biomolecular organization of ultra-structures in cells, has hindered its widest application in biomedical research. Here, in SMLM imaging of cellular structures such as lipid rafts and microtubules with saturation labelling, we identified ultra-bright localizations, each of which is contributed by simultaneous emission of multiple molecules within a diffraction-limit region and has been regarded before as a regular localization from single molecule. Consistently, ultra-bright localizations are also observed in simulated SMLM imaging of endoplasmic reticulum or microtubules from public resource. Furthermore, after calibrating each ultrabright localization into multiple single-molecule localizations using the photon-number-based models, the density of total localizations shows linear correlation with the true molecule density, presenting SMLM with new reconstruction method as a quantitative analysis approach. Therefore, identification and dissection of ultra-bright localizations in SMLM enable the close and quantitative estimate of the true biomolecular organization.

biophysics↗

bioRxiv2022

Generative pretraining from large-scale transcriptomes: Implications for single-cell deciphering and clinical translation

Exponential accumulation of single-cell transcriptomes poses great challenge for efficient assimilation. Here, we present an approach entitled tGPT towards integration of 22.3 million single-cell transcriptomes by modeling gene expression rankings as generative pretraining task. tGPT is conceptually simple in that it autoregressively models the ranking of a gene in the context of its preceding neighbors. We demonstrated the high performance of tGPT on a range of fundamental single-cell analysis tasks and novel applications on bulk tissues. The single-cell clusters and cell lineage trajectories derived from tGPT are highly aligned with known cell labels and states. The feature patterns of tumor bulk tissues learned by tGPT are associated with a wide range of genomic alteration events, prognosis and treatment outcome of immunotherapy. tGPT represents a new analytical paradigm for integrating and deciphering massive amount of transcriptome data and it will facilitate the interpretation and clinical translation of single-cell transcriptomes.

bioinformatics↗

bioRxiv2022

The CIC-ERF co-deletion underlies fusion independent activation of ETS family member, ETV1, to drive prostate cancer progression

The dysregulation of ETS family transcription factors drives human prostate cancer. The majority of prostate cancer is the result of chromosomal rearrangements that lead to aberrant ETS gene expression. The mechanisms that lead to fusion independent ETS factor upregulation and prostate oncogenesis remain unknown. Here, we show that two neighboring transcription factors, Capicua (CIC) and ETS2 repressor factor (ERF), which are co-deleted in human prostate tumors can drive prostate oncogenesis. Concurrent CIC and ERF loss commonly occurs through focal genomic deletions at chromosome 19q13.2. Mechanistically, CIC and ERF co-bind the proximal regulatory element and mutually repress the ETS transcription factor, ETV1. Targeting ETV1 in CIC and ERF deficient prostate cancer limits tumor growth. Thus, we have uncovered a fusion independent mode of ETS transcriptional activation defined by concurrent loss of CIC and ERF.

cancer biology↗

bioRxiv2021

Susceptibilities of human ACE2 genetic variants in coronavirus infection

The COVID-19 pandemic, caused by SARS-CoV-2, has resulted in more than 1603 million cases worldwide and 3.4 million deaths (as of May 2021), with varying incidences and death rates among regions/ethnicities. Human genetic variation can affect disease progression and outcome, but little is known about genetic risk factors for SARS-CoV-2 infection. The coronaviruses SARS-CoV, SARS-CoV-2 and HCoV-NL63 all utilize the human protein angiotensin-converting enzyme 2 (ACE2) as the receptor to enter cells. We hypothesized that the genetic variability in ACE2 may contribute to the variable clinical outcomes of COVID-19. To test this hypothesis, we first conducted an in silico investigation of single-nucleotide polymorphisms (SNPs) in the coding region of ACE2 gene. We then applied an integrated approach of genetics, biochemistry and virology to explore the capacity of select ACE2 variants to bind coronavirus spike protein and mediate viral entry. We identified the ACE2 D355N variant that restricts the spike protein-ACE2 interaction and consequently limits infection both in vitro and in vivo. In conclusion, ACE2 polymorphisms could modulate susceptibility to SARS-CoV-2, which may lead to variable disease severity.

microbiology↗

bioRxiv2021

Characterizing collaborative transcription regulation with a graph-based deep learning approach

Human epigenome and transcription activities have been characterized by a number of sequence-based deep learning approaches which only utilize the DNA sequences. However, transcription factors interact with each other, and their collaborative regulatory activities go beyond the linear DNA sequence. Therefore leveraging the informative 3D chromatin organization to investigate the collaborations among transcription factors is critical. We developed ECHO, a graph-based neural network, to predict chromatin features and characterize the collaboration among them by incorporating 3D chromatin organization from 200-bp high-resolution Micro-C contact maps. ECHO predicts 2, 583 chromatin features with significantly higher average AUROC and AUPR than the best sequence-based model. We observed that chromatin contacts of different distances affected different types of chromatin features prediction in diverse ways, suggesting complex and divergent collaborative regulatory mechanisms. Moreover, ECHO was interpretable via gradient-based attribution methods. The attributions on chromatin contacts identify important contacts relevant to chromatin features. The attributions on DNA sequences identify TF binding motifs and TF collaborative binding. Furthermore, combining the attributions on contacts and sequences reveals important sequence patterns in the neighborhood which are relevant to target sequences chromatin feature prediction. The attribution results that reveal TF collaboration activities are provided on a website https://echo.dcmb.med.umich.edu/echo/.

bioinformatics↗

bioRxiv2021

3'HS1 CTCF binding site in human β-globin locus regulates fetal hemoglobin expression

Mutations in the adult {beta}-globin gene can lead to a variety of hemoglobinopathies, including sickle cell disease and {beta}-thalassemia. An increase in fetal hemoglobin expression throughout adulthood, a condition named Hereditary Persistence of Fetal Hemoglobin (HPFH), has been found to ameliorate hemoglobinopathies. Deletional HPFH occurs through the excision of a significant portion of the 3 end of the {beta}-globin locus, including a CTCF binding site termed 3HS1. Here, we show that the deletion of this CTCF site alone induces fetal hemoglobin expression in both adult CD34+ hematopoietic stem and progenitor cells and HUDEP-2 erythroid progenitor cells. This induction is driven by the ectopic access of a previously postulated distal enhancer located in the OR52A1 gene downstream of the locus, which can also be insulated by the inversion of the 3HS1 CTCF site. This suggests that genetic editing of this binding site can have therapeutic implications to treat hemoglobinopathies.

genomics↗

bioRxiv2021

An Aerobic Anoxygenic Phototrophic Bacterium Fixes CO2 via the Calvin-Benson-Bassham Cycle

Aerobic anoxygenic phototrophic bacteria (AAnPB) are photoheterotrophs, which use light as auxiliary energy and require organic carbon (OC) for growth. Herein, we report the unusual strain B3, which is a true AAnPB because it requires oxygen for growth, harbours genes for cbb3- and bd-type cytochromes and acsF, and produces bacteriochlorophyll. The B3 genome encodes the complete metabolic pathways for AAnPB light utilization, CO2 fixation via Calvin-Benson-Bassham (CBB) cycle and oxidation of sulfite and H2, and the transcriptome indicated that all components of these pathways were fully transcribed. Expression of the marker genes related to photosynthesis, including pufM for light harnessing and rbcL for CO2 fixation, and the activity of RubisCO, the key enzyme in the Calvin-Benson-Bassham (CBB) cycle, increased in response to decreased OC supply. Large amounts of cell biomass were obtained in liquid BG11 medium under illumination. The strain thus likely photoautotrophically grows using sulfite or H2 as an electron donor. Similar GC contents between photosynthesis, the CBB cycle and 16S rRNA genes and the consistency of their phylogenetic topologies implied that light harnessing and carbon fixation genes evolved vertically from an anaerobic phototrophic ancestor of Rhodospirillaceae in Alphaproteobacteria. In conclusion, strain B3 represents a novel AAnPB characterized by photoautotrophy using the CBB cycle. This kind of AAnPB may be ecologically significant in the global carbon cycle.

microbiology↗

bioRxiv2021

Molecular basis for SPINDOC-Spindlin1 engagement and its role in transcriptional inhibition

Spindlin1 is a transcriptional coactivator with three Tudor-like domains, of which the first and second Tudors are engaged in histone methylation readout, while the function of the third Tudor is largely unknown. Recent studies revealed that the transcriptional co-activator activity of Spindlin1 could be attenuated by SPIN*DOC. Here we solved the crystal structure of SPIN*DOC-Spindlin1 complex, revealing that a hydrophobic motif, DOCpep3 (256-281), of SPIN*DOC interacts with Tudor 3 of Spindlin1 and completes its {beta}-barrel fold. Massive hydrophobic contacts and hydrogen bonding interactions ensure a high affinity DOCpep3-Spindlin1 engagement with a binding Kd of 30 nM. Interestingly, we characterized two more K/R-rich motifs of SPIN*DOC, DOCpep1 (187-195) and DOCpep2 (228-239), which bind to Spindlin1 at lower affinities with Kd values of 78 M and 31 M, respectively. Structural and binding studies revealed that DOCpep1 and DOCpep2 competitively bind to the aromatic cage of Spindlin1 Tudor 2 that is responsible for H3K4me3 readout. Although DOCpep3-Spindlin1 engagement is compatible with histone readout, an extended SPIN*DOC fragment containing DOCpep1 and DOCpep2 inhibits histone or TCF4 binding by Spindin1 due to introduced competition. This inhibitory effect is more pronounced for weaker binding targets but not for strong ones such as H3 "K4me3-K9me3" bivalent mark. Our RT-qPCR experiment showed that the removal of the hydrophobic motif or the K/R-rich region compromised the inhibitory effects of SPIN*DOC on Spindlin1-mediated transcriptional activation. In sum, here we revealed multivalent engagement between SPIN*DOC and Spindlin1, in which a hydrophobic motif acts as the primary binding site for stable SPIN*DOC-Spindlin1 association, while two more neighboring K/R-rich motifs further modulate the target selectivity of Spindlin1 via competitive inhibition, therefore attenuating the transcriptional co-activator activities of Spindlin1 through affecting its chromatin association.

biophysics↗