Baltica: integrated splice junction usage analysis

Alternative splicing is a tightly regulated co- and post-transcriptional process contributing to the transcriptome diversity observed in eukaryotes. Several methods for detecting differential junction usage (DJU) from RNA sequencing (RNA-seq) datasets exist. Yet, efforts to integrate the results from DJU methods are lacking. Here, we present Baltica, a framework that provides workflows for quality control, de novo transcriptome assembly with StringTie2, and currently 4 DJU methods: rMATS, JunctionSeq, Majiq, and LeafCutter. Baltica puts the results from different DJU methods into context by integrating the results at the junction level. We present Baltica using 2 datasets, one containing known artificial transcripts (SIRVs) and the second dataset of paired Illumina and Oxford Nanopore Technologies RNA-seq. The data integration allows the user to compare the performance of the tools and reveals that JunctionSeq outperforms the other methods, in terms of F1 score, for both datasets. Finally, we demonstrate for the first time that meta-classifiers trained on scores of multiple methods outperform classifiers trained on scores of a single method, emphasizing the application of our data integration approach for differential splicing identification. Baltica is available at https://github.com/dieterich-lab/Baltica under MIT license.

Identifying common transcriptome signatures of cancer by interpreting deep learning models

Cancer is a set of diseases characterized by unchecked cell proliferation and invasion of surrounding tissues. The many genes that have been genetically associated with cancer or shown to directly contribute to oncogenesis vary widely between tumor types, but common gene signatures that relate to core cancer pathways have also been identified, signifying that cancer cases display common hallmark molecular features. It is not clear however whether there exist additional sets of genes or transcriptomic features that are less well known in cancer biology but that are also commonly deregulated across several cancer types. Here, in order to agnostically identify transcriptomic features that are commonly shared between cancer types, we used RNA-Seq datasets encompassing thousands of samples from 19 healthy tissue types and 18 solid tumor types to train three feed-forward neural networks, based either on protein-coding gene expression, lncRNA expression or splice junction use, to distinguish between healthy and tumor samples. All three models achieve high precision, recall and accuracy on test sets derived from 13 datasets used during training and on an independent test dataset, indicating that our models recognize transcriptome signatures that are consistent across tumors. Analysis of attribution values extracted from our models reveals that genes that are commonly altered in cancer by expression or splicing variations are under strong evolutionary and selective constraints, suggesting that they have important cellular functions. Importantly, we found that genes composing our cancer transcriptome signatures are not frequently affected by mutations or genomic alterations and that their functions differ widely from the genes genetically associated with cancer. Finally, our results also highlighted that deregulation of RNA-processing genes and aberrant splicing are pervasive features across a large array of solid tumor types. The transcriptomic features that we highlight here define cancer signatures that may reflect causal variations or consequences of disease state, or a combination of both.

The Pan-Clk/Dyrk Inhibitor Cirtuvivint (SM08502) Exposes Mechanistic Underpinnings of Alternative Splicing As a Therapeutic Vulnerability in Heme Malignancies

Mutations in spliceosomal genes are one of the most common classes of somatic alterations in patients with Myelodysplastic Syndrome (MDS) and occur across the entire spectrum of myeloid malignancies, including 10‒25% of patients with acute myeloid leukemia (AML). These mutations occur in higher proportions in AML subjects greater than 60 years of age, or when AML has transformed from an antecedent MDS. Spliceosomal gene mutations are implicated in the production of pathological RNA splicing patterns that block cell differentiation and maintain a myeloid precursor phenotype. This suggests deranged pre-mRNA splicing is a mechanistic determinant of many heme malignancies and, as such, has provoked interest in therapeutic modulation of pre-mRNA splicing as a treatment paradigm. The CLK/DYRK family of protein kinases has been recognized as an integration hub for signal transduction-dependent modulation of alternative pre-mRNA splice junction selection through direct phosphorylation of the serine/arginine-rich splicing factor (SRSF) splice junction enhancer-binding proteins. Thus, these kinases potentially represent a druggable intervention point in alternative splicing-dependent cancers. The isoquinoline SM08502 (cirtuvivint) is a potent ATP-competitive inhibitor of the Cdc2-like kinases (CLK1-4) and the dual specificity tyrosine phosphorylation-regulated kinases (DYRK1-4) with activity against only a minimal number of the remaining members of the CMGC-family kinases and the kinome as a whole. Here the consequence of pan-CLK/DYRK kinase inhibition on cell viability and tumorigenicity was evaluated across a panel of human tumor-derived AML, DLBCL, MCL, myeloma, T-ALL, and CML/CLL models. EC 50s in response to cirtuvivint ranged from 0.014 μM‒0.495 μM in 4-day in vitro cell viability assays, with low-dose responders enriched in subsets of AML, myeloma, DLBCL, MCL and T-ALL. Cell viability EC 50s were associated with induction of programed cell death at drug exposures that inhibited accumulation of phosphorylated SRSF proteins and the anti-apoptotic protein MCL-1. To directly evaluate the contribution of CLK/DYRK kinases to alternative splicing profiles, high-depth RNAseq analysis was performed across 4 cell lines (3 acute myeloid leukemia cell lines and 1 mantle cell lymphoma cell line) +/- a 6-hour exposure to 1µM cirtuvivint. Both baseline and drug-induced changes in alternative splicing events (ASEs) were measured using a multivariate analysis of splicing transcripts (rMATS). The frequency of cirtuvivint-induced ASEs was approximately 20% of total detected ASEs. Concordant drug-induced ASEs among the tested cell lines were in genes enriched in pathways known to drive oncogenesis in hematopoietic lineages, including the MAP kinase and mTOR signaling pathways. Tumor growth inhibition assays in immunocompromised mice showed a range of model-specific responses, including tumor stasis and partial to complete tumor regression at clinically relevant exposures. A cell-based synthetic-lethal screen of cirtuvivint across 36 small molecule inhibitors identified multiple BCL-2 inhibitors among the most prominent synergistic combinations. Consistent with this, combination of the BCL-2 inhibitor venetoclax with cirtuvivint was sufficient to induce tumor regressions in AML xenograft models (KG-1 and HL-60) that were resistant to either single-agent drug at the same concentrations. These observations support further evaluation of CLK/DYRK inhibitors as a therapeutic strategy for heme malignancies dependent upon alternative pre-mRNA splicing. Disclosures Bossard: Biosplice Therapeutics: Current Employment. McMillan: Prizer: Ended employment in the past 24 months. Beaupre: Pfizer: Ended employment in the past 24 months. White: Pfizer: Ended employment in the past 24 months.

Extracellular circular RNA profiles in plasma and urine of healthy, male college athletes

Circular RNA (circRNA) are a recently discovered class of RNA characterized by a covalently-bonded back-splice junction. As circRNAs are inherently more stable than other RNA species, they may be detected extracellularly in peripheral biofluids and provide novel biomarkers. While circRNA have been identified previously in peripheral biofluids, there are few datasets for circRNA junctions from healthy controls. We collected 134 plasma and 114 urine samples from 54 healthy, male college athlete volunteers, and used RNASeq to determine circRNA content. The intersection of six bioinformatic tools identified 965 high-confidence, characteristic circRNA junctions in plasma and 72 in urine. Highly-expressed circRNA junctions were validated by qRT-PCR. Longitudinal samples were collected from a subset, demonstrating circRNA expression was stable over time. Lastly, the ratio of circular to linear transcripts was higher in plasma than urine. This study provides a valuable resource for characterization of circRNA in plasma and urine from healthy volunteers, one that can be developed and reassessed as researchers probe the circRNA contents of biofluids across physiological changes and disease states.

NanoSplicer: Accurate identification of splice junctions using Oxford Nanopore sequencing

Motivation: Long read sequencing methods have considerable advantages for characterising RNA isoforms. Oxford nanopore sequencing records changes in electrical current when nucleic acid traverses through a pore. However, basecalling of this raw signal (known as a squiggle) is error prone, making it challenging to accurately identify splice junctions. Existing strategies include utilising matched short-read data and/or annotated splice junctions to correct nanopore reads but add expense or limit junctions to known (incomplete) annotations. Therefore, a method that could accurately identify splice junctions solely from nanopore data would have numerous advantages. Results: We developed "NanoSplicer" to identify splice junctions using raw nanopore signal (squiggles). For each splice junction the observed squiggle is compared to candidate squiggles representing potential junctions to identify the correct candidate. Measuring squiggle similarity enables us to compute the probability of each candidate junction and find the most likely one. We tested our method using 1. synthetic mRNAs with known splice junctions 2. biological mRNAs from a lung-cancer cell-line. The results from both datasets demonstrate NanoSplicer improves splice junction identification, especially when the basecalling error rate near the splice junction is elevated. Our method is implemented in the software package NanoSplicer, available at https://github.com/shimlab/NanoSplicer.

CiLiQuant: Quantification of RNA Junction Reads Based on Their Circular or Linear Transcript Origin

Distinguishing circular RNA (circRNA) reads from reads derived from the linear host transcript is a challenging task because of sequence overlap. We developed a computational approach, CiLiQuant, that determines the relative circular and linear abundance of transcripts and gene loci using backsplice and forward splice junction reads generated by existing mapping and circRNA discovery tools.

Specific splice junction detection in single cells with SICILIAN

Precise splice junction calls are currently unavailable in scRNA-seq pipelines such as the 10x Chromium platform but are critical for understanding single-cell biology. Here, we introduce SICILIAN, a new method that assigns statistical confidence to splice junctions from a spliced aligner to improve precision. SICILIAN is a general method that can be applied to bulk or single-cell data, but has particular utility for single-cell analysis due to that data’s unique challenges and opportunities for discovery. SICILIAN’s precise splice detection achieves high accuracy on simulated data, improves concordance between matched single-cell and bulk datasets, and increases agreement between biological replicates. SICILIAN detects unannotated splicing in single cells, enabling the discovery of novel splicing regulation through single-cell analysis workflows.

Empirical prediction of variant-associated cryptic-donors with 87% sensitivity and 95% specificity

Predicting which cryptic-donors may be activated by a genetic variant is notoriously difficult. Through analysis of 5,145 cryptic-donors activated by 4,811 variants (versus 86,963 decoy-donors not used; any GT or GC), we define an empirical method predicting cryptic-donor activation with 87% sensitivity and 95% specificity. Strength (according to four algorithms) and proximity to the authentic-donor appear important determinants of cryptic-donor activation. However, other factors such as auxiliary splicing elements, which are difficult to identify, play an important role and are likely responsible for current prediction inaccuracies. We find that the most frequent mis-splicing events at each exon-intron junction, mined from 40,233 RNA-sequencing samples, predict with remarkable accuracy which cryptic-donor will be activated in rare disease. Aggregate RNA-Sequencing splice-junction data provides an accurate, evidence-based method to predict variant-activated cryptic-donors in genetic disorders, assisting pathology consideration of possible consequences of a variant for the encoded protein and RNA diagnostic testing strategies.

Empirical prediction of variant-associated cryptic-donors with 87% sensitivity and 95% specificity

Predicting which cryptic-donors may be activated by a genetic variant is notoriously difficult. Through analysis of 5,145 cryptic-donors activated by 4,811 variants (versus 86,963 decoy-donors not used; any GT or GC), we define an empirical method predicting cryptic-donor activation with 87% sensitivity and 95% specificity. Strength (according to four algorithms) and proximity to the authentic-donor appear important determinants of cryptic-donor activation. However, other factors such as auxiliary splicing elements, which are difficult to identify, play an important role and are likely responsible for current prediction inaccuracies. We find that the most frequent mis-splicing events at each exon-intron junction, mined from 40,233 RNA-sequencing samples, predict with remarkable accuracy which cryptic-donor will be activated in rare disease. Aggregate RNA-Sequencing splice-junction data provides an accurate, evidence-based method to predict variant-activated cryptic-donors in genetic disorders, assisting pathology consideration of possible consequences of a variant for the encoded protein and RNA diagnostic testing strategies.