Biology subjects

CHEN, X.

Publications and source records attributed to CHEN, X..

4 recordsLinked to original sources

bioRxiv2024

Subject-Agnostic Transformer-Based Neural Speech Decoding from Surface and Depth Electrode Signals

ObjectiveThis study investigates speech decoding from neural signals captured by intracranial electrodes. Most prior works can only work with electrodes on a 2D grid (i.e., Electrocorticographic or ECoG array) and data from a single patient. We aim to design a deep-learning model architecture that can accommodate both surface (ECoG) and depth (stereotactic EEG or sEEG) electrodes. The architecture should allow training on data from multiple participants with large variability in electrode placements and the trained model should perform well on participants unseen during training. ApproachWe propose a novel transformer-based model architecture named SwinTW that can work with arbitrarily positioned electrodes by leveraging their 3D locations on the cortex rather than their positions on a 2D grid. We train subject-specific models using data from a single participant and multi-patient models exploiting data from multiple participants. Main ResultsThe subject-specific models using only low-density 8x8 ECoG data achieved high decoding Pearson Correlation Coefficient with ground truth spectrogram (PCC=0.817), over N=43 participants, outperforming our prior convolutional ResNet model and the 3D Swin transformer model. Incorporating additional strip, depth, and grid electrodes available in each participant (N=39) led to further improvement (PCC=0.838). For participants with only sEEG electrodes (N=9), subject-specific models still enjoy comparable performance with an average PCC=0.798. The multi-subject models achieved high performance on unseen participants, with an average PCC=0.765 in leave-one-out cross-validation. SignificanceThe proposed SwinTW decoder enables future speech neuropros-theses to utilize any electrode placement that is clinically optimal or feasible for a particular participant, including using only depth electrodes, which are more routinely implanted in chronic neurosurgical procedures. Importantly, the generalizability of the multi-patient models suggests that such a model can be applied to new patients that do not have paired acoustic and neural data, providing an advance in neuroprostheses for people with speech disability, where acoustic-neural training data is not feasible.

neuroscience↗

bioRxiv2023

Cross-Dataset Identification of Human Disease-Specific Cell Subtypes Enabled by the Gene Print-based Algorithm--gPRINT

Despite extensive efforts in developing cell annotation algorithms for single cell RNA sequencing results, most algorithms fail to achieve cross-dataset mapping of cell subtypes due to factors such as batch effects between datasets. This limitation is particularly evident when rapidly annotating disease-specific cell subtypes across multiple datasets. In this study, we present gPRINT, a machine learning tool that utilizes the unique one-dimensional "gene print" expression patterns of individual cells. gPRINT is capable of automatically predicting cell types and annotating disease-specific cell subtypes. The development of gPRINT involved curation and harmonization of public datasets, algorithm validation within and across datasets, and the annotation of disease-specific fibroblast subtypes across various disease subgroups and datasets. Additionally, we created a preliminary single-cell atlas of human tendinopathy fibroblasts and successfully achieved automatic prediction of disease-specific cell subtypes in tendon disease. Furthermore, we conducted an exploration of key targets and related drugs specific to this subtype in tendon disease. The proposed approach offers an automated and unified method for identifying disease-specific cell subtypes across datasets, serving as a valuable reference for annotating fibroblast-specific subtypes in different disease states and facilitating the exploration of therapeutic targets in tendon disease.

bioinformatics↗

bioRxiv2023

CaMKII autophosphorylation but not downstream kinase activity is required for synaptic memory

CaMKII plays a critical role in long-term potentiation (LTP), a well-established model for learning and memory through the enhancement of synaptic transmission. Biochemical studies indicate that CaMKII catalyzes a phosphotransferase (kinase) reaction of both itself (autophosphorylation) and of multiple downstream target proteins. However, whether either type of phosphorylation plays any role in the synaptic enhancing action of CaMKII remains hotly contested. We have designed a series of experiments to define the minimal requirements for the synaptic enhancement by CaMKII. We find that autophosphorylation of T286 and further binding of CaMKII to the GluN2B subunit are required both for initiating LTP and for its maintenance (synaptic memory). Once bound to the NMDA receptor, the synaptic action of CaMKII occurs in the absence of kinase activity. Thus, autophosphorylation, together with binding to the GluN2B subunit, are the only two requirements for CaMKII in synaptic memory.

neuroscience↗

bioRxiv2019

Single-Cell RNA-Seq Reveals Novel Mitochondria-related Musculoskeletal Cell Populations during Adult Axolotl Limb Regeneration Process

While the capacity to regenerate tissues or limbs is limited in mammals including humans, unlike us, axolotls are able to regrow entire limbs and major organs. The wound blastema have been extensively studied in limb regeneration. However, due to the inadequate characterization and coordination of cell subpopulations involved in the regeneration process, it hinders the discovery of the key clue for human limb regeneration. In this study, we applied unbiased large-scale single-cell RNA sequencing to classify cells throughout the adult axolotl limb regeneration process. We computationally identified 7 clusters in regenerating limbs, including the novel regeneration-specific mitochondria-related cluster supporting regeneration through energy providing and the COL2+ cluster contributing to regeneration through cell-cell interactions signals. We also discovered the dedifferentiation and re-differentiation of the COL1+/COL2+ cellular subpopulation and uncovered a COL2-mitochondria sub-cluster supporting the musculoskeletal system regeneration. On the basis of these findings, we reconstructed the dynamic single-cell transcriptome atlas of adult axolotl limb regenerative process, and identified the novel regenerative mitochondria-related musculoskeletal populations, which yielded deeper insights into the crucial interactions between cell clusters within the regenerative microenvironment.

genomics↗