Small RNA-Seq and profiling Y. Hoogstrate 1,2 1 Department of Bioinformatics & Department of Urology ErasmusMC, Rotterdam 2 CTMM Translational Research IT (TraIT) BioSB: 5th RNA-seq data analysis course, Leiden
Overview Small RNA-Seq - primary target: micrornas Characterization of the Melanoma mirnaome by Deep Sequencing Beyond micrornas
Main application of Small RNA-Seq profiling MicroRNAs (mirnas) 18 23bp [4]
Main application of Small RNA-Seq profiling MicroRNAs (mirnas) Mature Mature-star [2]
Main application of Small RNA-Seq profiling MicroRNAs (mirnas) Mature Mature-star mornas [1]
Obtaining small RNA-seq data (Illumina.com) No fragmentation Size selection Read size often 30bp or 35bp Part of primer / adapters present in most reads! Often stranded protocol
Characterization of the Melanoma mirnaome by Deep Sequencing Case study: Characterization of mirnas Characterization of the Melanoma mirnaome by Deep Sequencing[9] 12 samples (human pigment cells) Goal: find existing + novel mirnas mirna expression profiles to see if samples group by cell type (validation)
Characterization of the Melanoma mirnaome by Deep Sequencing MiRNA characterization workflow (miranalyzer) QA/QC Adapter removal (very important) If mirna=22bp and read=30bp, 8 bases are adapter Trimming low-q bases? I would not do this, because this might predict shorter mirnas Profiling Alignment Spliced alignment? (mirtrons, some trnas; 14-60 nt) Align to: reference genome or known mirnas (e.g. mirbase) Reference genome / transcriptome if you want to find novel Max multimaps? Minimum read count? Predict novel mirnas
Characterization of the Melanoma mirnaome by Deep Sequencing Profiling novel mirnas (miranalyzer) [3] Followed by classification, including features based on Secondary structure of pre-mirna Expression / read count Alignment
Characterization of the Melanoma mirnaome by Deep Sequencing Profiling novel mirnas - quanitification
Characterization of the Melanoma mirnaome by Deep Sequencing Profiling novel mirnas - quanitification
Characterization of the Melanoma mirnaome by Deep Sequencing Profiling novel mirnas - quanitification
Small RNA-Seq - primary target: micrornas Beyond micrornas Characterization of the Melanoma mirnaome by Deep Sequencing Profiling novel mirnas [9] I Count reads of all mirnas I Validation: check if clustering confirms biological subtypes RNA-Seq analysis in Galaxy 8 September 2015 References
Characterization of the Melanoma mirnaome by Deep Sequencing Next steps with a full repertoire of mirnas Differential mirna expression analysis Similar to RNA-Seq (use pre-mirnas or allow offset) Order on gene count, en see if Let-7 is at the top Be careful with multi-maps (many homologue mirnas) mirna target analysis Discover targets: e.g. Correlation analysis with mrna-seq data Discover targets: find matches in genome Use known targets: Gene onthology and text mining
Beyond micrornas: RNA in pieces many different RNA fragments derived from small RNA species other than microrna [8] [10]
Beyond micrornas: characterizing all small RNAs Adapter removal Aligning with non-splicing aligners (bowtie, bwa, etc.) Except for a few trnas and mirtrons, small RNAs are usually not spliced Alignment to: 1 Database of small RNAs of interest, e.g. mirbase no novel mirnas etc. 2 Reference genome Will introduce more multi-map reads 3 First 1, then 2, then merge More complicated, but solves both issues Reads are small, changes on multi-map are high, use stranded protocol if possible
Beyond micrornas: characterizing all small RNAs
Beyond micrornas: RNA in pieces QC: alignment - read counts per host [7]
Analysis: detection of novel small RNAs Classical strategy 1: predict mirna folding (energy) + overlapping reads X Very specific for mirnas [2]
Analysis: detection of novel small RNAs Classical strategy 2: derive normal distribution from read density V Focus on mornas (not 2D-structure specific) X Assumes read density over small RNA to be symmetrical X If variance is not correct, predicted small RNAs may become huge - A small RNA is usually cleft at 5 and 3 independently: two independent events [6]
Analysis: detection of novel small RNAs Detect both 5 and 3 individually and join back together (FlaiMapper) [5]
FlaiMapper on Human dataset 1 Adapter removal 2 Align reads to database of all annotated ncrnas 3 Detect novel small RNAs Classify by type of precursor (pre-mirna, snorna, trna, etc.) All types of ncrnas seem to produce small RNAs
Novel small ncrnas All types of ncrnas seem to produce small RNAs [7]
Novel small ncrnas QC: length distribution of predicted small RNAs [7]
Small RNA-Seq workflow 1 Adapter removal 2 Align reads to database of all annotated ncrnas 3 Detect novel small RNAs QC: length distribution of predicted small RNAs 4 Quantify expression levels (known + detected novel small RNAs) Because of the small size of ncrnas, add an offset of 3 5bp to annotations on both sides Be careful with conclusions: many multi-mappers in mirnas, snorna- and trna (many homologues) 5 Diffential gene expression analysis, identical in classical RNA-seq, (EdgeR, DeSeq, CLC Bio, etc.) 6 Interpretation of DE small RNAs
Interpretation of DE small RNAs Sounds intuitive If small RNA is up- or down-regulated, relation to precursor is important (trf up, trna also up?) mirnas expression in relation to target mrnas novel Small RNAs: location in precursor (2D structure?) Little is known about Small RNA derived fragments (their roles, their function, their structure, etc.)
SnoRNAs [7] Multiple and overlapping sdrnas Multiple degradation mechanisms? Correlation sdrnas from same host snorna? Correlation with host snorna itself?
References I [1] Stefania Bortoluzzi, Marta Biasiolo, and Andrea Bisognin. Micrornaoffset {RNAs} (mornas): by-product spectators or functional players? Trends in Molecular Medicine, 17(9):473 474, 2011. [2] Marc R. Friedlander, Wei Chen, Catherine Adamidi, Jonas Maaskola, Ralf Einspanier, Signe Knespel, and Nikolaus Rajewsky. Discovering micrornas from deep sequencing data using mirdeep. Nat Biotech, 26(4):407 415, Apr 2008. [3] Michael Hackenberg, Naiara Rodrguez-Ezpeleta, and Ana M. Aransay. miranalyzer: an update on the detection and analysis of micrornas in high-throughput sequencing experiments. Nucleic Acids Research, 39(suppl 2):W132 W138, 2011. [4] Lin He and Gregory J. Hannon. Micrornas: small rnas with a big role in gene regulation. Nat Rev Genet, 5(7):522 531, Jul 2004. [5] Youri Hoogstrate, Guido Jenster, and Elena S. Martens-Uzunova. Flaimapper: computational annotation of small ncrna-derived fragments using rna-seq high-throughput data. Bioinformatics, 31(5):665 673, 2015. [6] David Langenberger, Clara Bermudez-Santana, Jana Hertel, Steve Hoffmann, Philipp Khaitovich, and Peter F. Stadler. Evidence for human microrna-offset rnas in small rna sequencing data. Bioinformatics, 25(18):2298 2301, 2009. [7] Elena Martens-Uzunova, Youri Hoogstrate, Anton Kalsbeek, Bas Pigmans, Mirella den Berg, Natasja Dits, Sren Nielsen, Adam Baker, Tapio Visakorpi, Chris Bangma, and Guido Jenster. C/d-box snorna-derived rna production is associated with malignant transformation and metastatic progression in prostate cancer. Oncotarget, 6(19), 2015. [8] Elena S. Martens-Uzunova, Michael Olvedy, and Guido Jenster. Beyond microrna novel {RNAs} derived from small non-coding {RNA} and their implication in cancer. Cancer Letters, 340(2):201 211, 2013. Next Generation Sequencing Applications in Cancer Research. [9] Mitchell S. Stark, Sonika Tyagi, Derek J. Nancarrow, Glen M. Boyle, Anthony L. Cook, David C. Whiteman, Peter G. Parsons, Christopher Schmidt, Richard A. Sturm, and Nicholas K. Hayward. Characterization of the melanoma mirnaome by deep sequencing. PLoS ONE, 5(3):e9685, 03 2010. [10] Alex C. Tuck and David Tollervey. {RNA} in pieces. Trends in Genetics, 27(10):422 432, 2011.