TLGP: a flexible transfer learning algorithm for gene prioritization based on heterogeneous source domain

Imputation for Single-cell RNA-seq Data with Non-negative Matrix Factorization and Transfer Learning

Single-cell RNA sequencing (scRNA-seq) has been proven to be an effective technology for investigating the heterogeneity and transcriptome dynamics due to the single-cell resolution. However, one of the major problems for data obtained by scRNA-seq is excessive zeros in the count matrix, which hinders the downstream analysis enormously. Here, we present a method that integrates non-negative matrix factorization and transfer learning (NMFTL) to impute the scRNA-seq data. It borrows gene expression information from the additional dataset and adds graph-regularized terms to the decomposed matrices. These strategies not only maintain the intrinsic geometrical structure of the data itself but also further improve the accuracy of estimating the expression values by adding the transfer term in the model. The real data analysis result demonstrates that the proposed method outperforms the existing matrix-factorization-based imputation methods in recovering dropout entries, preserving gene-to-gene and cell-to-cell relationships, and in the downstream analysis, such as cell clustering analysis, the proposed method also has a good performance. For convenience, we have implemented the "NMFTL" method with R scripts, which could be available at https://github.com/FocusPaka/NMFTL.

Cell Features Reconstruction from Gene Association Network of Single Cell

Gene expression as an unstable form of cell characterization has been widely used for single-cell analyses. Although there are cell-specific networks (CSN) to explore stable gene associations within a single cell, the amount of information in CSN is huge and there is no method to measure the interaction level between genes. Therefore, this paper presents a two-level approach to reconstructing single-cell features, which transforms the original gene expression feature into the gene ontology feature and gene interaction feature. Specifically, we first squeeze all CSNs into a cell network feature matrix (CNFM) by fusing the global position and neighborhood influence of genes. Next, we propose a computational method of gene gravitation based on CNFM to quantify the extent of gene-gene interaction, and we can construct a gene gravitation network for single cells. Finally, we further design a novel index of gene gravitation entropy to quantitatively evaluate the level of single-cell differentiation. The experiments on eight different scRNA-seq datasets show the effectiveness and broad application prospects of our method.

Hierarchical shared transfer learning for biomedical named entity recognition

BACKGROUND

Biomedical named entity recognition (BioNER) is a basic and important medical information extraction task to extract medical entities with special meaning from medical texts. In recent years, deep learning has become the main research direction of BioNER due to its excellent data-driven context coding ability. However, in BioNER task, deep learning has the problem of poor generalization and instability.

RESULTS

we propose the hierarchical shared transfer learning, which combines multi-task learning and fine-tuning, and realizes the multi-level information fusion between the underlying entity features and the upper data features. We select 14 datasets containing 4 types of entities for training and evaluate the model. The experimental results showed that the F1-scores of the five gold standard datasets BC5CDR-chemical, BC5CDR-disease, BC2GM, BC4CHEMD, NCBI-disease and LINNAEUS were increased by 0.57, 0.90, 0.42, 0.77, 0.98 and - 2.16 compared to the single-task XLNet-CRF model. BC5CDR-chemical, BC5CDR-disease and BC4CHEMD achieved state-of-the-art results.The reasons why LINNAEUS's multi-task results are lower than single-task results are discussed at the dataset level.

CONCLUSION

Compared with using multi-task learning and fine-tuning alone, the model has more accurate recognition ability of medical entities, and has higher generalization and stability.