Transcriptomics: RNA-Seq

Data Analysis

For details see Martin and Wang (2011).

Transcriptome assembly falls into three main categories

1. Reference-based strategy: Requires the availability of a reference genome.
1. RNA-Seq reads are aligned to a reference genome with a splice-aware aligner.
2. Overlapping reads from each locus are clustered to build a graph representing all possible isoforms.
3. The graph is traversed to identify individual isoforms.
2. De novo strategy: does not require a reference genome; uses de Bruin graph theory (see Compeau et al, 2011) and leverages the redundancy between short read sequences.
1. For a set of k-mers from each read, de novo assemblers look for overlapping regions of length k-1 in the k-mers. K-mers are connected if there is exactly a k-1 overlap, and a de Bruijn graph is constructed.
2. Connected k-mers are collapsed to form longer sequences until a divergence in the graph is reached based on sequences from a different read. The graph may rejoin or diverge multiple times before all the k-mers re collapsed into a summarizing graph.
3. Multiple isoforms are assembled by traversing the collapsed graph.
3. Combined strategy: Uses both the reference-based and de novo strategies.

Assembled transcripts are post-processed to remove assembly errors and the expression level of each transcript is estimated by counting the number of reads that align to the transcript

• 1
• 2
• 3
• 4

• Data Quality Control »