International Journal of Data Mining and Bioinformatics (3 papers in press)

Regular Issues

• Nonlinear Optimization with Constraints for Four-parameter Logistic Model used in Cell-based In-vitro Assay  
  by Shan Chen, Tianhong Pan, Haoran Li 
  Abstract: Cell-based in-vitro assays are commonly used to perform chemical toxicity assessments. These assays usually employ a technical replicate to increase the reliability of the experimental data. Realizing consistent assessment using the replicates is a key challenge in cell-based in-vitro assays. In this study, a novel constrained nonlinear optimization that estimates the four-parameter logistic (4PL) model is proposed to overcome variability in the replicate measurements. First, the tested substances toxicity intensities are calculated by comparing its time-dependent cellular response curves (TCRCs) with the TCRC of the negative control, which evaluates the cell inhibition/death at a particular time point. Next, the variability of each toxicity intensity is set as a discount factor, and a constrained nonlinear optimization is constructed. The LevenbergMarquardt algorithm is used to obtain the optimal parameters of the 4PL model. Furthermore, a linearized 4PL model is presented to set initialized values for nonlinear optimization. Two case studies are conducted to validate the proposed method. The analyzed results confirm that the proposed method achieves consistent results.
  Keywords: Four-parameter logistic (4PL) model; cytotoxicity; Real-time cell analyzer (RTCA); nonlinear optimization.
  
  
• Dynamic basis of mRNA processing in male germline stem cell pluripotency and reprogramming by alternative polyadenylation  
  by Praveen Kumar Guttula, Mukesh Kumar Gupta 
  Abstract: Male germline stem (GS) cells are responsible for sperm production in males throughout the adulthood. Under appropriate in vitro conditions, these cells can also undergo the reprogramming events to generate germline pluripotent stem (GPS) cells. However, the mechanism of reprogramming of GS cells to GPS cells is elusive. This study investigated the dynamics of mRNA processing in GS and GPS cells and involvement of alternative polyadenylation (APA) in pluripotency and reprogramming using a novel computational approach for high throughput mRNA and miRNA datasets. The APA usage pattern in GS and GPS cells was identified in microarray gene expression datasets using custom java files, and was analyzed for 3' Untranslated Regions (3'-UTR) length in differentially expressed mRNAs, Gene Ontology, and Protein Interactions Networks by using various computational tools. The upstream and downstream regions of polyadenylation sites (PAS) were further analyzed for cis-elements and motifs. In addition, the long- and short- isoforms of APA-regulated genes were predicted for their possible regulation by miRNAs. Results showed that, APA events regulated the expression of 78 genes and caused the shortening of 3'-UTR during the reprogramming of GS cells to GPS cells. The APA-regulated genes were found to be involved in the regulation of mRNA processing and RNA splicing with shortened 3'-UTR due to proximal PAS. The proximal PAS also had a high occurrence of TC-rich cis-elements, which recruits CstF complexes to regulate cell proliferation and differentiation in GPS cells. Importantly, 28 miRNAs were found to target the APA-regulated genes and were present in their 3'-UTR region. These results suggest that APA events might coordinate the reprogramming of GS cells to GPS cells by dynamic processing of 3'-UTR length in differentially expressed mRNAs through miRNAs.
  Keywords: alternate polyadenylation; germline stem cells; germline pluripotent stem cells; reprogramming; microRNA; polyadenylation site.
  
  
• HNC: A hybrid neighborhood-consensus clustering algorithm for single-cell RNA-seq data  
  by Priyojit Das, Sujay Saha 
  Abstract: With the advent of single-cell RNA-seq (scRNA-seq) technology, the study of transcriptomic activity at single cell resolution has become extremely popular among the researchers. Analysis of the expression data generated from scRNA-seq technique has the potential to reveal unprecedented information about the heterogeneity in gene expression among tissue cells during both fixed point and development time. Although numerous statistical methods are used to analyze single cell transcriptomes, still pretty much area is open there to design analysis methods for the cell type identification. In this paper, a hybrid neighborhood-consensus (HNC) clustering algorithm is proposed to identify cellular states from single-cell gene expression data. The hybrid algorithm transforms the original dataset by combining k-nearest neighbor adjacency matrix and consensus matrix obtained from single-cell expression matrix and then uses modified k-means algorithm to cluster transformed dataset. To compare the performance of the proposed HNC algorithm with other unsupervised clustering methods, we used 12 real scRNA-seq datasets (cell types include - cancer, embryonic, pancreatic, lung and renal cell). From the comparison result, it is found that the HNC algorithm outperforms other standard single-cell analysis methods in terms of three external cluster validation indexes - Adjusted Rand Index, Purity and Normalized Mutual Information.
  Keywords: Single-cell RNA sequencing; Clustering; Consensus clustering; Neighborhood graph; Embryonic development; Clonal heterogeneity.