Guitar2: An improved version of the Guitar for sketching the transcriptomic view of RNA-related biological features.

Biological elements, such as genes, exons, coding sequences, are usually expressed as genomic features based on genome-based coordinates. However, the RNA transcription landmarks are usually expressed in the form of RNA-based coordinates. To analyze the association between RNA-related genomic features and RNA transcription landmarks, some tools, such as Guitar, have been developed to convert between these two coordinate systems. However, there remain some issues, such as incomplete transcriptional structures, limitation of transcriptomic view analysis, etc. Therefore, we made corresponding improvements based on Guitar, considered the promoter, 5' cap, and 3' poly-A tail structure in the transcript, standardized the input format of RNA-related genomic features, and finally developed Guitar2. Guitar2 converts genome-based coordinates and RNA-based coordinates with a more accurate strategy. Besides, Guitar2 supports the sketching of three different transcriptional views based on the overall transcripts, messenger RNA, as well as long non-coding RNA. The analysis of m6A modification using Guitar2 shows that m6A modification is significantly enriched near the stop codon in the mRNA, which is consistent with the known results. In conclusion, Guitar2's improvement of coordinate system structure and the provision of full transcriptional view will contribute to the further analysis of RNA-related biological features. Guitar2 is now publicly available from R/Bioconductor: https://bioconductor.org/packages/release/bioc/html/Guitar.html.

Learning Siamese networks for laser vision seam tracking.

To design a stable laser vision seam-tracking system, an advanced weld image processing algorithm based on Siamese networks is investigated and proposed to resist the interference of arc and spatter in the welding process. This specially designed neural network, combined with powerful feature expression capabilities of deep learning, takes two welding images with different sizes as inputs and generates a target confidence map in a single forward pass by using the cross-correlation algorithm. To prevent the error accumulation and model drift, an online update strategy via local cosine similarity is developed. The use of metal inert-gas welding can realize real-time and precious tracking under the condition that the strong arc continuously shields the welding seam feature points.

FGFICA: Independent Component Analysis of Fusion Genomic Features for Mining Epi-Transcriptome Profiling Data.

Existing studies indicate that in-depth studies of the N6-methyladenosine (m6A) co-methylation patterns in epi-transcriptome profiling data may contribute to understanding its complex regulatory mechanisms. In order to fully utilize the potential features of epi-transcriptome data and consider the advantages of independent component analysis (ICA) in local pattern mining tasks, we propose an ICA algorithm that fuses genomic features (FGFICA) to discover potential functional patterns. FGFICA first extracts and fuses the confidence information, homologous information, and genomic features implied in epi-transcriptome profiling data and then solves the model based on negative entropy maximization. Finally, to mine m6A co-methylation patterns, the probability density of the extracted independent components is estimated. In the experiment, FGFICA extracted 64 m6A co-methylation patterns from our collected MeRIP-seq high-throughput data. Further analysis of some selected patterns revealed that the m6A sites involved in these patterns were highly correlated with four m6A methylases, and these patterns were significantly enriched in some pathways known to be regulated by m6A.