Accurate Identification and Analysis of Human mRNA Isoforms Using Deep Long Read Sequencing

Hagen Tilgner, Debasish Raha, Lukas Habegger, Mohammed Mohi-Ud-Din, Mark Gerstein, M SnyderG3 2013 · DOI

TL;DR

This paper demonstrates the power of deep long-read sequencing (specifically using the 454 platform) for accurate human transcriptome analysis. By generating millions of long reads, the researchers could identify novel gene structures, analyze alternative splicing in an annotation-free manner, and reconstruct full-length transcripts more effectively than with short-read data alone.

What Problem Does This Paper Address?

Accurately identifying and quantifying all the different RNA isoforms produced by a genome (the transcriptome) is a fundamental challenge in biology. Current methods primarily rely on short-read sequencing (like Illumina), where short DNA fragments are sequenced, aligned to a reference genome, and then computationally assembled into potential transcripts. This approach has limitations, especially for complex transcript structures with many introns, as short reads often don't span entire genes or even multiple introns. This makes it difficult to fully capture the diversity of transcript isoforms, particularly for less-studied genes or in newly sequenced genomes lacking comprehensive annotations. The paper aims to show how long-read sequencing can overcome these limitations.

Methods

The core of the study involves using the 454/Roche sequencing platform to generate long complementary DNA (cDNA) reads from human K562 and HeLa S3 cell lines. Key methodological aspects include:

  • Deep Sequencing: Generating millions of reads (over 4 million per cell line) with a median length exceeding 500 bp.
  • Annotation-Independent Mapping: Aligning these long reads to the human genome (hg19) using the GMAP aligner without relying on pre-existing gene annotations. This allows for the discovery of novel gene structures.
  • High-Confidence Mapping Criteria: Developing specific criteria (mapLength and mapQuality) to filter for reliable alignments, ensuring that a significant portion of the read matches the genome accurately.
  • Alternative Splicing Analysis: Developing methods to identify and quantify alternative splicing events, including inclusion/exclusion levels of exons, in an annotation-free manner. This involved defining metrics like "inclusion level" (Ψ-value).
  • Transcript Reconstruction: Using the Cufflinks software to assemble full-length transcripts from the long-read alignments.

Datasets

  • Human K562 and HeLa S3 Cell Lines: Two large datasets of long-read cDNA sequences were generated from these human cell lines.
  • ENCODE Project Data: The study also references and compares its findings with de novo transcriptome predictions generated by the ENCODE project using short-read RNA-seq data.

Key Insights and Ideas

  • Long Reads Uncover the Unseen: The ability of long reads to span multiple exons and introns is crucial for identifying novel transcript structures that are missed by short-read sequencing.
  • Annotation-Free Analysis is Powerful: The study demonstrates that robust transcriptome analysis, including alternative splicing, can be performed without relying on existing gene annotations, reducing potential biases.
  • Complementary Technologies: Long-read sequencing is not a replacement for short-read sequencing but a powerful complement. Short reads excel at deep quantitative analysis, while long reads provide structural insights.
  • lncRNAs are a Rich Source of Novelty: Long-read sequencing is particularly effective at revealing the complex structures of long non-coding RNAs (lncRNAs), a class of genes that has only recently gained significant attention.

Key Results

  • High Mapping Rate: Over 94% of the generated long reads could be mapped to the human genome.
  • Novel Gene Structures: Approximately 15% of the gene structures identified by long-read mapping were not found in previous de novo analyses of short-read data.
  • lncRNA Discovery: A significant fraction of novel gene structures were identified for long non-coding RNA (lncRNA) genes.
  • Accurate Alternative Splicing: Cell type-specific alternative splicing could be reliably analyzed in an annotation-free manner, with 97% of identified alternative exons corresponding to known exons.
  • Full-Length Transcript Assembly: Long reads facilitated the prediction of ~9500 full-length transcripts, with over 50% of these having identical intron structures to annotated transcripts, a much higher rate than achieved with short-read data.
  • Improved Annotation: The study suggests that long-read sequencing can significantly improve existing genome annotations by revealing previously uncharacterized transcript isoforms.

Limitations and Future Work

  • Sequencing Depth vs. Length: While the study generated millions of reads, the depth was still less than what is achievable with short-read technologies. This limits the ability to perform highly quantitative expression analysis.
  • Platform Specificity: The study primarily used the 454 platform. The authors note that advancements in other long-read technologies (like PacBio and Oxford Nanopore) and improvements in short-read technologies (longer read lengths) will further enhance transcriptome analysis.
  • Cost: The cost of long-read sequencing was a consideration, and the authors suggest coupling it with capture techniques for genes with low expression.
  • Future Work: The authors envision that combining long-read and short-read sequencing will provide the most comprehensive and accurate transcriptome descriptions. They also highlight the potential for long-read sequencing in annotating newly sequenced genomes.

Summary updated

Abstract

Precise identification of RNA-coding regions and transcriptomes of eukaryotes is a significant problem in biology. Currently, eukaryote transcriptomes are analyzed using deep short-read sequencing experiments of complementary DNAs. The resulting short-reads are then aligned against a genome and annotated junctions to infer biological meaning. Here we use long-read complementary DNA datasets for the analysis of a eukaryotic transcriptome and generate two large datasets in the human K562 and HeLa S3 cell lines. Both data sets comprised at least 4 million reads and had median read lengths greater than 500 bp. We show that annotation-independent alignments of these reads provide partial gene structures that are very much in-line with annotated gene structures, 15% of which have not been obtained in a previous de novo analysis of short reads. For long-noncoding RNAs (i.e., lncRNA) genes, however, we find an increased fraction of novel gene structures among our alignments. Other important aspects of transcriptome analysis, such as the description of cell type-specific splicing, can be performed in an accurate, reliable and completely annotation-free manner, making it ideal for the analysis of transcriptomes of newly sequenced genomes. Furthermore, we demonstrate that long read sequence can be assembled into full-length transcripts with considerable success. Our method is applicable to all long read sequencing technologies.
BibTeX
@inproceedings{tilgner2013accurate,
  author = {Hagen Tilgner and Debasish Raha and Lukas Habegger and Mohammed Mohi-Ud-Din and Mark Gerstein and M Snyder},
  title = {{Accurate Identification and Analysis of Human mRNA Isoforms Using Deep Long Read Sequencing}},
  booktitle = {G3},
  year = {2013},
  doi = {10.1534/g3.112.004812},
}
Figure