2 Papers and their corresponding sub-goals
Table of contents
This thesis includes five papers, and each paper has a corresponding sub-goal (Table 2.1). This chapter gives a brief description of paper in context of each sub-goal.
Table 2.1. Papers and corresponding sub-goals of our research.
| Sub-goal | Paper No. | Title |
|---|---|---|
| miRNA target prediction | Paper 1 | MicroRNAs - targeting and target prediction |
| Paper 2 | A two step site and mRNA-level model for predicting microRNA targets | |
| Paper 5 | Inferring causative variants in microRNA target sites | |
| miRNA high-throughput experiments | Paper 3 | Target gene expression levels and competition between transfected and endogenous microRNAs are strong factors in high-throughput experiments |
| miRNA and other ncRNAs | Paper 4 | MicroRNAs affect gene expression by targeting cis-transcribed non-coding RNAs |
2.1 Three papers for the first sub-goal: miRNA target prediction
Paper 1: MicroRNAs - targeting and target prediction. This review paper outlines the features associated with animal miRNA targeting. It summarizes the characteristics of the features in six different categories: miRNA:mRNA paring, Site location, Conservation, Site accessibility, Multiple sites, and Expression profiles. It also contains a list of 30 different miRNA target prediction tools with information of feature coverage in context of the six categories.
Paper 2: A two step site and mRNA-level model for predicting microRNA targets. This paper presents a miRNA target prediction model that recognizes both the individual characteristics of functional binding sites and the global characteristics of miRNA-targeted mRNAs. Our novel two-step SVM model trains site level features at the first step, and, subsequently, it trains mRNA level features at the second step. Benchmark experiments showed that our two-step SVM model had a higher overall performance than other established miRNA target prediction tools.
Paper 5: Inferring causative variants in microRNA target sites. This paper shows an example that miRNA predictions from our two-step SVM model performs better than the other prediction algorithms when the predictions are used by other tools. Laurent F. Thomas was the main contributor to this study, and he developed a tool that can help identifying Single-nucleotide polymorphisms (SNPs) associated with diseases by focusing on SNPs affecting miRNA regulation. The tool uses miRNA target predictions to check the influence of SNPs that affect miRNA targeting. It can use any miRNA prediction tools that generate scores of miRNA target predictions. The paper showed that the tool had the best performance when our two-step SVM model was used.
2.2 One paper for the second sub-goal: miRNA high-throughput experiments
Paper 3: Target gene expression levels and competition between transfected and endogenous microRNAs are strong confounding factors in microRNA high-throughput experiments. This paper shows characteristics of different miRNA high-throughput experiments. Analysis on these high-throughput experiment data sometimes show inconsistent miRNA regulation factors, for example, one experiment shows 3’ UTR length is one of the most important factors, whereas other experiment shows it is least important. We investigated several factors that might affect this inconsistence, and we revealed that competition between endogenous miRNAs and the ectopically expressed miRNAs significantly contributed to the differences among different miRNA high-throughput experiments. We also found that this competition effect affected other factors, such as mRNA expression level and 3’ UTR length, in terms of miRNA targeting.
2.3 One paper for the third sub-goal: miRNA and other ncRNAs
Paper 4: MicroRNAs affect gene expression by targeting cis-transcribed non-coding RNAs. This paper shows potential miRNA regulation on two types of complex loci: cis-natural antisense transcripts (cis-NATs) and chromatin associated RNAs (CARs). We used several different types of data from high-throughput miRNA experiments to infer potential miRNA regulation on such loci. Our statistical analyses revealed that complex loci containing non-coding cis-NATs or CARs appeared to be under strong regulation, although this type of miRNA targeting is less prevalent than miRNA targeting of 3’ UTRs.