Comprehensive identification of diverse ribosomal RNA modifications by targeted nanopore direct RNA sequencing and JACUSA2.

Ribosomal RNAs are decorated by numerous post-transcriptional modifications whose exact roles in ribosome biogenesis, function, and human pathophysiology remain largely unknown. Here, we report a targeted direct rRNA sequencing approach involving a substrate selection step and demonstrate its suitability to identify differential modification sites in combination with the JACUSA2 software. We compared JACUSA2 to other tools designed for RNA modification detection and show that JACUSA2 outperforms other software with regard to detection of base modifications such as methylation, acetylation and aminocarboxypropylation. To illustrate its widespread usability, we applied our method to a collection of CRISPR-Cas9 engineered colon carcinoma cells lacking specific enzymatic activities responsible for particular rRNA modifications and systematically compared them to isogenic wild-type RNAs. Besides the numerous 2'-O methylated riboses and pseudouridylated residues, our approach was suitable to reliably identify differential base methylation and acetylation events. Importantly, our method does not require any prior knowledge of modification sites or the need to train complex models. We further report for the first time detection of human rRNA modifications by direct RNA-sequencing on Flongle flow cells, the smallest-scale nanopore flow cell available to date. The use of these smaller flow cells reduces RNA input requirements, making our workflow suitable for the analysis of samples with limited availability and clinical work.

Mapping of RNA Modifications by Direct Nanopore Sequencing and JACUSA2.

RNA modifications exist in all kingdom of life. Several different types of base or ribose modifications are now summarized under the term "epitranscriptome." With the advent of high-throughput sequencing technologies, much progress has been made in understanding RNA modification biology and how these modifications can influence many aspects of RNA life. The most widespread internal modification on mRNA is m6A, which has been implicated in physiological processes as well as disease pathogenesis. Here, we provide a workflow for the mapping of m6A sites using Nanopore direct RNA sequencing data. Our strategy employs pairwise comparison of basecalling error profiles with JACUSA2. We outline a general strategy for RNA modification detection on mRNA and describe two specific use cases on m6A detection in detail. Use case 1: a sample of interest with modifications (e.g., "wild-type" sample) is compared to a sample lacking a specific modification type (e.g., "knockout" sample, here METTL3-KO) or Use case 2: a sample of interest with modifications is compared to a sample lacking all modifications (e.g., in vitro transcribed cDNA). We provide a detailed protocol on experimental and computational aspects. Extensive online material provides a snakemake pipeline to identify m6A positions in mRNA and to validate the results against a miCLIP-derived m6A reference set. The general strategy is flexible and can be easily adapted by users in different application scenarios.

PathME: pathway based multi-modal sparse autoencoders for clustering of patient-level multi-omics data.

BACKGROUND: Recent years have witnessed an increasing interest in multi-omics data, because these data allow for better understanding complex diseases such as cancer on a molecular system level. In addition, multi-omics data increase the chance to robustly identify molecular patient sub-groups and hence open the door towards a better personalized treatment of diseases. Several methods have been proposed for unsupervised clustering of multi-omics data. However, a number of challenges remain, such as the magnitude of features and the large difference in dimensionality across different omics data sources. RESULTS: We propose a multi-modal sparse denoising autoencoder framework coupled with sparse non-negative matrix factorization to robustly cluster patients based on multi-omics data. The proposed model specifically leverages pathway information to effectively reduce the dimensionality of omics data into a pathway and patient specific score profile. In consequence, our method allows us to understand, which pathway is a feature of which particular patient cluster. Moreover, recently proposed machine learning techniques allow us to disentangle the specific impact of each individual omics feature on a pathway score. We applied our method to cluster patients in several cancer datasets using gene expression, miRNA expression, DNA methylation and CNVs, demonstrating the possibility to obtain biologically plausible disease subtypes characterized by specific molecular features. Comparison against several competing methods showed a competitive clustering performance. In addition, post-hoc analysis of somatic mutations and clinical data provided supporting evidence and interpretation of the identified clusters. CONCLUSIONS: Our suggested multi-modal sparse denoising autoencoder approach allows for an effective and interpretable integration of multi-omics data on pathway level while addressing the high dimensional character of omics data. Patient specific pathway score profiles derived from our model allow for a robust identification of disease subgroups.