DeepGSEA: explainable deep gene set enrichment analysis for single-cell transcriptomic data.

MOTIVATION: Gene set enrichment (GSE) analysis allows for an interpretation of gene expression through pre-defined gene set databases and is a critical step in understanding different phenotypes. With the rapid development of single-cell RNA sequencing (scRNA-seq) technology, GSE analysis can be performed on fine-grained gene expression data to gain a nuanced understanding of phenotypes of interest. However, with the cellular heterogeneity in single-cell gene profiles, current statistical GSE analysis methods sometimes fail to identify enriched gene sets. Meanwhile, deep learning has gained traction in applications like clustering and trajectory inference in single-cell studies due to its prowess in capturing complex data patterns. However, its use in GSE analysis remains limited, due to interpretability challenges. RESULTS: In this paper, we present DeepGSEA, an explainable deep gene set enrichment analysis approach which leverages the expressiveness of interpretable, prototype-based neural networks to provide an in-depth analysis of GSE. DeepGSEA learns the ability to capture GSE information through our designed classification tasks, and significance tests can be performed on each gene set, enabling the identification of enriched sets. The underlying distribution of a gene set learned by DeepGSEA can be explicitly visualized using the encoded cell and cellular prototype embeddings. We demonstrate the performance of DeepGSEA over commonly used GSE analysis methods by examining their sensitivity and specificity with four simulation studies. In addition, we test our model on three real scRNA-seq datasets and illustrate the interpretability of DeepGSEA by showing how its results can be explained. AVAILABILITY AND IMPLEMENTATION: https://github.com/Teddy-XiongGZ/DeepGSEA.

ProtoCell4P: an explainable prototype-based neural network for patient classification using single-cell RNA-seq.

MOTIVATION: The rapid advance in single-cell RNA sequencing (scRNA-seq) technology over the past decade has provided a rich resource of gene expression profiles of single cells measured on patients, facilitating the study of many biological questions at the single-cell level. One intriguing research is to study the single cells which play critical roles in the phenotypes of patients, which has the potential to identify those cells and genes driving the disease phenotypes. To this end, deep learning models are expected to well encode the single-cell information and achieve precise prediction of patients' phenotypes using scRNA-seq data. However, we are facing critical challenges in designing deep learning models for classifying patient samples due to (i) the samples collected in the same dataset contain a variable number of cells-some samples might only have hundreds of cells sequenced while others could have thousands of cells, and (ii) the number of samples available is typically small and the expression profile of each cell is noisy and extremely high-dimensional. Moreover, the black-box nature of existing deep learning models makes it difficult for the researchers to interpret the models and extract useful knowledge from them. RESULTS: We propose a prototype-based and cell-informed model for patient phenotype classification, termed ProtoCell4P, that can alleviate problems of the sample scarcity and the diverse number of cells by leveraging the cell knowledge with representatives of cells (called prototypes), and precisely classify the patients by adaptively incorporating information from different cells. Moreover, this classification process can be explicitly interpreted by identifying the key cells for decision making and by further summarizing the knowledge of cell types to unravel the biological nature of the classification. Our approach is explainable at the single-cell resolution which can identify the key cells in each patient's classification. The experimental results demonstrate that our proposed method can effectively deal with patient classifications using single-cell data and outperforms the existing approaches. Furthermore, our approach is able to uncover the association between cell types and biological classes of interest from a data-driven perspective. AVAILABILITY AND IMPLEMENTATION: https://github.com/Teddy-XiongGZ/ProtoCell4P.

Interpretable meta-learning of multi-omics data for survival analysis and pathway enrichment.

MOTIVATION: Despite the success of recent machine learning algorithms' applications to survival analysis, their black-box nature hinders interpretability, which is arguably the most important aspect. Similarly, multi-omics data integration for survival analysis is often constrained by the underlying relationships and correlations that are rarely well understood. The goal of this work is to alleviate the interpretability problem in machine learning approaches for survival analysis and also demonstrate how multi-omics data integration improves survival analysis and pathway enrichment. We use meta-learning, a machine-learning algorithm that is trained on a variety of related datasets and allows quick adaptations to new tasks, to perform survival analysis and pathway enrichment on pan-cancer datasets. In recent machine learning research, meta-learning has been effectively used for knowledge transfer among multiple related datasets. RESULTS: We use meta-learning with Cox hazard loss to show that the integration of TCGA pan-cancer data increases the performance of survival analysis. We also apply advanced model interpretability method called DeepLIFT (Deep Learning Important FeaTures) to show different sets of enriched pathways for multi-omics and transcriptomics data. Our results show that multi-omics cancer survival analysis enhances performance compared with using transcriptomics or clinical data alone. Additionally, we show a correlation between variable importance assignment from DeepLIFT and gene coenrichment, suggesting that genes with higher and similar contribution scores are more likely to be enriched together in the same enrichment sets. AVAILABILITY AND IMPLEMENTATION: https://github.com/berkuva/TCGA-omics-integration.

Turbidity-tolerant underwater wireless optical communications using dense blue-green wavelength division multiplexing.

Underwater wireless optical communication (UWOC) has demonstrated high-speed and low-latency properties in clear and coastal ocean water because of the relatively low attenuation 'window' for blue-green wavelengths from 450 nm to 550 nm. However, there are different attenuation coefficients for transmission in ocean water at different wavelengths, and the light transmission more seriously deteriorates with fluctuations in the water turbidity. Therefore, traditional UWOC using a single wavelength or coarse blue-green wavelengths has difficulty tolerating variations in water turbidity. Dense wavelength division multiplexing (WDM) technology provides sufficient communication channels with a narrow wavelength spacing and minimal channel crosstalk. Here, we improve the UWOC in clear and coastal ocean water using dense blue-green WDM. A cost-effective WDM emitter is proposed with directly modulated blue-green laser diodes. Dense wavelength beam combination and collimation are demonstrated in a 20-metre underwater channel from 490 nm to 520 nm. Demultiplexing with a minimum channel spacing of 2 nm is realized by an optical grating. Remarkably, our WDM results demonstrate an aggregate data rate exceeding 10 Gbit/s under diverse water turbidity conditions, with negligible crosstalk observed for each channel. This is the densest WDM implementation with a record channel spacing of 2 nm and the highest channel count for underwater blue-green light communications, providing turbidity-tolerant signal transmission in clear and coastal ocean water.

Design, synthesis and activity evaluation of pseudilin analogs against cyanobacteria as IspD inhibitors.

The discovery of safe, effective, and selective chemical algicides is the stringent need for the algicides development, and it is also one of the effective routes to control cyanobacteria harmful algal blooms and to meet the higher requirements of environmental and ecological. In this work, a series of novel bromo-N-phenyl-5-o-hydroxyphenylpyrazole-3-carboxyamides were rationally designed as pseudilin analogs by bioisosteric replacement and molecular hybridization strategies, in which the pyrrole unit of pseudilin was replaced with pyrazole and further combined with the dominant structural fragments of algicide diuron. The synthesis was carried out by a facile four-step routeincluding cyclization, amidation, transanulation, and halogenation. The biological activity evaluation on AtIspD, EcIspD, Synechocystis sp. PCC6803 and Microcystis aeruginosa FACHB905 revealed that most compounds had good EcIspD and excellent cyanobacteria inhibitory activity. In particular, compound 6bb exhibited potent algicidal activity against PCC6803 and FACHB905 with EC50 = 1.28 µM and 0.37 µM, respectively, 1.4-fold and 4.0-fold enhancement compared to copper sulfate (EC50 = 1.79 and 1.49 µM, respectively), and it also showed the best inhibitory activity of EcIspD. The binding of 6bb to EcIspD was explored by molecular docking, and it was confirmed that 6bb could bind to the EcIspD active site. Compound 6bb was proven to be a potential structure for the further development of novel algicides that targets IspD in the MEP pathway.

Fine Mapping of Candidate Gene Controlling Anthocyanin Biosynthesis for Purple Peel in Solanum melongena L.

Fruit color is an intuitive quality of horticultural crops that can be used as an evaluation criterion for fruit ripening and is an important factor affecting consumers' purchase choices. In this study, a genetic population from the cross of green peel 'Qidong' and purple peel '8 guo' revealed that the purple to green color of eggplant peel is dominant and controlled by a pair of alleles. Bulked segregant analysis (BSA), SNP haplotyping, and fine genetic mapping delimited candidate genes to a 350 kb region of eggplant chromosome 10 flanked by markers KA2381 and CA8828. One ANS gene (EGP22363) was predicted to be a candidate gene based on gene annotation and sequence alignment of the 350-kb region. Sequence analysis revealed that a single base mutation of 'T' to 'C' on the exon green peel, which caused hydrophobicity to become hydrophilic serine, led to a change in the three-level spatial structure. Additionally, EGP22363 was more highly expressed in purple peels than in green peels. Collectively, EGP22363 is a strong candidate gene for anthocyanin biosynthesis in purple eggplant peels. These results provide important information for molecular marker-assisted selection in eggplants, and a basis for analyzing the regulatory pathways responsible for anthocyanin biosynthesis in eggplants.

Continual knowledge infusion into pre-trained biomedical language models.

MOTIVATION: Biomedical language models produce meaningful concept representations that are useful for a variety of biomedical natural language processing (bioNLP) applications such as named entity recognition, relationship extraction and question answering. Recent research trends have shown that the contextualized language models (e.g. BioBERT, BioELMo) possess tremendous representational power and are able to achieve impressive accuracy gains. However, these models are still unable to learn high-quality representations for concepts with low context information (i.e. rare words). Infusing the complementary information from knowledge-bases (KBs) is likely to be helpful when the corpus-specific information is insufficient to learn robust representations. Moreover, as the biomedical domain contains numerous KBs, it is imperative to develop approaches that can integrate the KBs in a continual fashion. RESULTS: We propose a new representation learning approach that progressively fuses the semantic information from multiple KBs into the pretrained biomedical language models. Since most of the KBs in the biomedical domain are expressed as parent-child hierarchies, we choose to model the hierarchical KBs and propose a new knowledge modeling strategy that encodes their topological properties at a granular level. Moreover, the proposed continual learning technique efficiently updates the concepts representations to accommodate the new knowledge while preserving the memory efficiency of contextualized language models. Altogether, the proposed approach generates knowledge-powered embeddings with high fidelity and learning efficiency. Extensive experiments conducted on bioNLP tasks validate the efficacy of the proposed approach and demonstrates its capability in generating robust concept representations.

Highly-time-resolved FMCW LiDAR with synchronously-nonlinearity-corrected acquisition for dynamic locomotion.

Highly-time-resolved and precise tracking of position, velocity, and acceleration is urgently required when highly dynamic legged robots are walking, trotting, and jumping. Frequency-modulated continuous-wave (FMCW) laser ranging is able to provide precise measurement in short distance. However, FMCW light detection and ranging (LiDAR) suffers from a low acquisition rate and poor linearity of laser frequency modulation in wide bandwidth. A sub-millisecond-scale acquisition rate and nonlinearity correction in the wide frequency modulation bandwidth have not been reported in previous studies. This study presents the synchronous nonlinearity correction for a highly-time-resolved FMCW LiDAR. The acquisition rate of 20 kHz is obtained by synchronizing the measurement signal and the modulation signal of laser injection current with a symmetrical triangular waveform. The linearization of laser frequency modulation is conducted by resampling of 1000 intervals interpolated in every up-sweep and down-sweep of 25 µs, while measurement signal is stretched or compressed in every period of 50 µs. The acquisition rate is demonstrated to be equal to the repetition frequency of laser injection current for the first time to the best of authors' knowledge. This LiDAR is successfully used to track the foot trajectory of a jumping single-leg robot. The high velocity up to 7.15 m/s and high acceleration of 365 m/s2 are measured during the up-jumping phase, while heavy shock takes place with high acceleration of 302 m/s2 as the foot end strikes the ground. The measured foot acceleration of over 300 m/s2, which is more than 30 times gravity acceleration, is reported on a jumping single-leg robot for the first time.

Multifunctional water-based metamaterial with polarization conversion and absorption.

A multifunctional metamaterial to realize broadband x-to-y cross-polarization conversion (CPC) and the absorption of electromagnetic waves is proposed in this paper. The presented multifunctional water-based metamaterial (MWM) consists of the top metallic dielectric substrate, the middle 3D printed container, and the bottom metal backplane. When the container is filled with water, the polarization conversion ratio (PCR) reaches more than 90% at 5.8-9.4 GHz, and the excellent absorption performance is achieved in the frequency band of 16.1-16.9 GHz. In addition, the CPC is achieved in two frequency bands (5.9-10.0 GHz and 14.3-16.4 GHz) without water injection. The unique properties of the proposed design are validated by experiments. As expected, the MWM simultaneously achieves polarization conversion and absorption functions, which is meaningful and consequential for multifunction and conformal stealth applications.

Leaky-Vivaldi antenna covered with metasurface with leaky wave radiation and aperture radiation.

A leaky-Vivaldi antenna covered with metasurface (LVAM) is proposed in this paper. The traditional Vivaldi antenna covered with metasurface realizes backward frequency beam-scanning from -41∘ to 0∘ in the high-frequency operating band (HFOB) and retains aperture radiation in the low-frequency operating band (LFOB). In the LFOB, the metasurface can be regarded as a transmission line to realize a slow-wave transmission. In the HFOB, the metasurface can be considered a 2D periodic leaky-wave structure to realize a fast-wave transmission. The simulated results show that LVAM has the -10 dB return loss bandwidths of 46.5% and 40.0%, and the realized gain of 8.8-9.6 dBi and 11.8-15.2 dBi cover the 5 G Sub-6 GHz band (3.3-5.3 GHz) and the X band (8.0-12.0 GHz), respectively. The test results are in good agreement with the simulated results. As a dual-band antenna covering the 5 G Sub-6 GHz communication band and military radar band, the proposed antenna can guide the future integrated design of communication and radar antenna systems.

Design, Synthesis and Bioactivity Evaluation of Heterocycle-Containing Mono- and Bisphosphonic Acid Compounds.

Fosmidomycin (FOS) is a naturally occurring compound active against the 1-deoxy-D-xylulose 5-phosphate reductoisomerase (DXR) enzyme in the 2-C-methyl-D-erythritol 4-phosphate (MEP) pathway, and using it as a template for lead structure design is an effective strategy to develop new active compounds. In this work, by replacing the hydroxamate unit of FOS with pyrazole, isoxazole and the related heterocycles that also have metal ion binding affinity, while retaining the monophosphonic acid in FOS or replacing it with a bisphosphonic acid group, heterocycle-containing mono- and bisphosphonic acid compounds as FOS analogs were designed. The key steps involved in the facile synthesis of these FOS analogs included the Michael addition of diethyl vinylphosphonate or tetraethyl vinylidenebisphosphonate to ß-dicarbonyl compounds and the subsequent cyclic condensation with hydrazine or hydroxylamine. Two additional isoxazolinone-bearing FOS analogs were synthesized via the Michaelis-Becker reaction with diethyl phosphite as a key step. The bioactivity evaluation on model plants demonstrated that several compounds have better herbicidal activities compared to FOS, with the most active compound showing a 3.7-fold inhibitory activity on Arabidopsis thaliana, while on the roots and stalks of Brassica napus L. and Echinochloa crus-galli in a pre-emergence inhibitory activity test, the activities of this compound were found to be 3.2- and 14.3-fold and 5.4- and 9.4-fold, respectively, and in a post-emergency activity test on Amaranthus retroflexus and Echinochloa crus-galli, 2.2- and 2.0-fold inhibition activities were displayed. Despite the significant herbicidal activity, this compound exhibited a DXR inhibitory activity lower than that of FOS but comparable to that of other non-hydroxamate DXR inhibitors, and the dimethylallyl pyrophosphate rescue assay gave no statistical significance, suggesting that a different target might be involved in the inhibiting process. This work demonstrates that using bioisosteric replacement can be considered as a valuable strategy to discover new FOS analogs that may have high herbicidal activities.

Continual representation learning for evolving biomedical bipartite networks.

MOTIVATION: Many real-world biomedical interactions such as 'gene-disease', 'disease-symptom' and 'drug-target' are modeled as a bipartite network structure. Learning meaningful representations for such networks is a fundamental problem in the research area of Network Representation Learning (NRL). NRL approaches aim to translate the network structure into low-dimensional vector representations that are useful to a variety of biomedical applications. Despite significant advances, the existing approaches still have certain limitations. First, a majority of these approaches do not model the unique topological properties of bipartite networks. Consequently, their straightforward application to the bipartite graphs yields unsatisfactory results. Second, the existing approaches typically learn representations from static networks. This is limiting for the biomedical bipartite networks that evolve at a rapid pace, and thus necessitate the development of approaches that can update the representations in an online fashion. RESULTS: In this research, we propose a novel representation learning approach that accurately preserves the intricate bipartite structure, and efficiently updates the node representations. Specifically, we design a customized autoencoder that captures the proximity relationship between nodes participating in the bipartite bicliques (2 × 2 sub-graph), while preserving both the global and local structures. Moreover, the proposed structure-preserving technique is carefully interleaved with the central tenets of continual machine learning to design an incremental learning strategy that updates the node representations in an online manner. Taken together, the proposed approach produces meaningful representations with high fidelity and computational efficiency. Extensive experiments conducted on several biomedical bipartite networks validate the effectiveness and rationality of the proposed approach.

Embeddings of genomic region sets capture rich biological associations in lower dimensions.

MOTIVATION: Genomic region sets summarize functional genomics data and define locations of interest in the genome such as regulatory regions or transcription factor binding sites. The number of publicly available region sets has increased dramatically, leading to challenges in data analysis. RESULTS: We propose a new method to represent genomic region sets as vectors, or embeddings, using an adapted word2vec approach. We compared our approach to two simpler methods based on interval unions or term frequency-inverse document frequency and evaluated the methods in three ways: First, by classifying the cell line, antibody or tissue type of the region set; second, by assessing whether similarity among embeddings can reflect simulated random perturbations of genomic regions; and third, by testing robustness of the proposed representations to different signal thresholds for calling peaks. Our word2vec-based region set embeddings reduce dimensionality from more than a hundred thousand to 100 without significant loss in classification performance. The vector representation could identify cell line, antibody and tissue type with over 90% accuracy. We also found that the vectors could quantitatively summarize simulated random perturbations to region sets and are more robust to subsampling the data derived from different peak calling thresholds. Our evaluations demonstrate that the vectors retain useful biological information in relatively lower-dimensional spaces. We propose that vector representation of region sets is a promising approach for efficient analysis of genomic region data. AVAILABILITY AND IMPLEMENTATION: https://github.com/databio/regionset-embedding. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Towards underwater coherent optical wireless communications using a simplified detection scheme.

In this paper, an effective method of underwater coherent optical wireless communication (UCOWC) with a simplified detection scheme is proposed. The proof-of-concept experiments with M-ary PSK have been conducted with a common laser used for the signal source and local oscillator (LO). The BER performance has been evaluated at different underwater channel attenuations and the maximum achievable attenuation length (AL) with a BER below the forward error correction (FEC) limit of 3.8×10-3 is investigated. The tested system offers data rates of 500 Mbps, 1 Gbps, and 1.5 Gbps with the BPSK, QPSK and 8PSK modulated signals, respectively. The corresponding maximum achievable attenuation lengths are measured as 13.4 AL 12.5 AL, and 10.7 AL. In addition, the performance degradation of the practical system with separate free running lasers for the signal and LO is also estimated. To the best of our knowledge, the UCOWC system is proposed and experimentally studied for the first time. This work provides a simple and effective approach to take advantages of coherent detection in underwater wireless optical communication, opening a promising path toward the development of practical UCOWC with next-generation underwater data transmission requirements on the capacity and transmission distance.

Liquid-cladded optical phased array for a single-wavelength beam steering.

We report a simple concept to implement a single-wavelength beam steering based on a liquid-cladded one-dimensional (1D) optical phased array (OPA). The beam steering was realized by modifying the waveguide mode effective index through replacing the liquid upper claddings. A prototype of a 32-channel liquid-cladded OPA was fabricated and characterized. Owing to the high refractive index range of liquids (>0.625), a maximum steering angle of >10∘ was achieved with the liquid range from 1.0 to 1.63 at a wavelength of 940 nm. Moreover, the liquid-cladded OPA reveals a quasi-continuous beam steering range of >29∘ by combining the liquid cladding tuning and discrete wavelength tuning of λ=785nm, 852 nm, and 940 nm. Further integration with optofluidic systems offers the OPA potential for low power consumption and all-fluidic beam steering operating at a single wavelength.

Broadband silicon nitride nanophotonic phased arrays for wide-angle beam steering.

In this Letter, the broadband operation in wavelengths from 520 nm to 980 nm is demonstrated on silicon nitride nanophotonic phased arrays. The widest beam steering angle of 65° on a silicon nitride phased array is achieved. The optical radiation efficiency of the main grating lobe in a broad wavelength range is measured and analyzed theoretically. The optical spots radiated from the phased array chip are studied at different wavelengths of lasers. The nanophotonic phased array is excited by a supercontinuum laser source for a wide range of beam steering for the first time to the best of our knowledge. It paves the way to tune the wavelength from visible to near infrared range for silicon nitride nanophotonic phased arrays.

Parallel emitted silicon nitride nanophotonic phased arrays for two-dimensional beam steering.

In this Letter, a two-dimensional (2D) beam steering on silicon nitride (SiNx) nanophotonic phased arrays from visible to near-infrared wavelengths is reported for the first time, to the best of our knowledge. In order to implement beam steering along the transverse direction for one-dimensional waveguide surface grating arrays, wavelengths from 650 to 980 nm provided by the supercontinuum laser are used to excite the phased array. Then the beams are parallel radiated with steering angles in a sequence of 26.84° to -16.54∘ along the transverse direction, and a continuous line in the far field consisting of parallel emitted spots is produced with a total view angle of 43.38°. Moreover, this continuous far-field line is steered along the longitudinal direction with massive wavelengths simultaneously tuned by phase shifts from -π/2 to over +π/2. This method with massive parallel wavelengths emitted paves a new way for 2D steering on SiNx nanophotonic phased arrays.

MeSHProbeNet: a self-attentive probe net for MeSH indexing.

MOTIVATION: MEDLINE is the primary bibliographic database maintained by National Library of Medicine (NLM). MEDLINE citations are indexed with Medical Subject Headings (MeSH), which is a controlled vocabulary curated by the NLM experts. This greatly facilitates the applications of biomedical research and knowledge discovery. Currently, MeSH indexing is manually performed by human experts. To reduce the time and monetary cost associated with manual annotation, many automatic MeSH indexing systems have been proposed to assist manual annotation, including DeepMeSH and NLM's official model Medical Text Indexer (MTI). However, the existing models usually rely on the intermediate results of other models and suffer from efficiency issues. We propose an end-to-end framework, MeSHProbeNet (formerly named as xgx), which utilizes deep learning and self-attentive MeSH probes to index MeSH terms. Each MeSH probe enables the model to extract one specific aspect of biomedical knowledge from an input article, thus comprehensive biomedical information can be extracted with different MeSH probes and interpretability can be achieved at word level. MeSH terms are finally recommended with a unified classifier, making MeSHProbeNet both time efficient and space efficient. RESULTS: MeSHProbeNet won the first place in the latest batch of Task A in the 2018 BioASQ challenge. The result on the last test set of the challenge is reported in this paper. Compared with other state-of-the-art models, such as MTI and DeepMeSH, MeSHProbeNet achieves the highest scores in all the F-measures, including Example Based F-Measure, Macro F-Measure, Micro F-Measure, Hierarchical F-Measure and Lowest Common Ancestor F-measure. We also intuitively show how MeSHProbeNet is able to extract comprehensive biomedical knowledge from an input article.

Over 10 attenuation length gigabits per second underwater wireless optical communication using a silicon photomultiplier (SiPM) based receiver.

Underwater wireless optical communication (UWOC) will play an important role in the underwater environment exploration and marine resource development due to its advantages of high data rate and good mobility. However, the significant signal power attenuation in the underwater channel limits the transmission distance of UWOC. Attenuation length (AL) is widely used as an indicator for evaluating the UWOC system's long-distance transmission capability. At present, Gbps UWOC is limited within 7AL. Using a SiPM based receiver can dramatically increase the AL that UWOC can support. In this paper, a novel UWOC receiver built from an off-the-shelf SiPM has been demonstrated. The finite pulse width and limited bandwidth of SiPM limit the SiPM based UWOC system's data rate. To boost the system's data rate, an optimum method to process the SiPM's signal has therefore been investigated. Based on these methods, the communication capabilities of the SiPM based UWOC have been investigated experimentally. Results show that the SiPM based receiver can support 11.6AL without turbulence and 9.28AL within weak turbulence (scintillation index = 0.0447) at 1 Gbps.

Engineered Paramagnetic Graphene Quantum Dots with Enhanced Relaxivity for Tumor Imaging.

Nano contrast agents (Nano CA) are nanomaterials used to increase contrast in the medical magnetic resonance imaging (MRI). However, the related relaxation mechanism of the Nano CA is not clear yet and little significant breakthrough in relaxivity enhancement has been achieved. Herein, a new hydrophilic Gd-DOTA complex functionalized with different chain length of PEG was synthesized and incorporated into graphene quantum dots (GQD) to obtain paramagnetic graphene quantum dots (PGQD). We performed a variable-temperature and variable-field intensity NMR study in aqueous solution on the water exchange and rotational dynamics of three different chain lengths of PGQD. The optimal GQD with paramagnetic chain length shows a great improvement in performance on 1H NMR relaxometric studies. In vitro results demonstrated that the relaxivity of the designed PGQD could be controlled by regulating the PEG length, and its relaxivity was ∼16 times higher than that of current commercial MRI contrast agents (e.g., Gd-DTPA), on a "per Gd" basis. The relaxivity of the Nano CA can be rationally tuned to obtain unmatched potentials in MR imaging, exemplified by preparation of the paramagnetic GQD with the enhanced T1 relaxivity. The fabricated PGQDs with suitable PEG length got the best relaxivity at 1.5 T. After intravenous injection, its feeding process by solid tumor could even be monitored by clinically used 1.5 T MRI scanners. This research will also provide an excellent platform for the design and synthesis of highly effective MR contrast agents.