Extend rat gene models with CAGEscan

Background

• Some rat gene models miss the real 5′ ends.
• HelicosCAGE and CAGEscan libraries are available from a “universal” rat RNA preparation.

Goals

Contribute experimental evidence that extends and update the gene models in rat.

Key questions:

• How many rat “refseq” TSS’es are correct?
• How many rat “refseq” TSS’es are shifted, either 5’ or 3’?
• Function of shifted genes, ontological enrichment of shifted genes compared to the correctly annotated ones?
• How many new genes with support(x)?
• Investigate high confidence TSSs (evidence from both CAGE-methods)
• Alternative promoters
• Other

Project outline

• Setup of scripting repo
• Get an overview of data
• (Mapping), processing, filtering
• Descriptive analysis(How far away (both upstream and downstream) are the refseq TSS's from CAGE clusters and the other way around. I will do this for the provisional clusters(Mette)
• How to deal with triplicates. Variance filtering?
• Tag clustering
• Aggregation of existing gene models.
• Define support(x)
• Integration of existing rat RNA-seq data from SRA/GEO?

Definitions

• Correct annotation. Find the number of correct annotated genes/transcripts based on more datasources for instance UCSC genes, refseq genes and mRNAs. Set a cutoff for how close the cluster should be to a known TSS based on the descriptive analysis(Probably around +/-50).
• Alternative promoter: If we have CAGE evidence for both the known promoter and a new one close by there is probably an alternative promoter. But if there is no evidence for the known one the TSS is probably wrongly annotated.
• Definition to belong to a known promoter: Max shift allowed(see correct annotation)? The 3' cagescan tag falls in the exon of a known gene?
• Filter: Seen with both Cagescan and hCage. This could be part of the clustering.

Data

CAGEscan

Sample preparation and sequencing

10009-101B8 is the same RNA as used for CNhs10614, the ‘Universal RNA - Rat Normal Tissues’ HelicosCAGE library.

• NCig10012: 2 × 54 bp CAGEscan library, 6,903,269 reads. Index sequence GCTCAG.
• NCig10071: 2 × 36 bp experimental CAGEscan, 28,936,176 reads. Index sequences ACAGATGCTATA, ATCGTGGCTATA, CACGATGCTATA, CACTGAGCTATA, CTGACGGCTATA, GAGTGAGCTATA, GTATACGCTATA, TCGAGCGCTATA.
• NChi10001: 2 × 51 bp CAGEscan library (HiSeq test run), 9,662,576 reads. Index sequence GCTCAG.

Bzipped FASTQ files are available in <https://fantom5-collaboration.gsc.riken.jp/webdav/home/Rat_cagescan/FASTQ/>. See CAGEscan_mapping_protocol on what to trim from the reads before aligning.

Name scheme: name_lane_direction.fq.bz2. The sequencer lane is indicated but should not have importance. Direction 1 is 5′ and direction 2 is 3′.

FASTQ sequences of sorted and trimmed reads can be retreived from the paired-end aligned BAM files (see below), that include unmapped reads.

Mapping (rn4)

Available at: https://fantom5-collaboration.gsc.riken.jp/webdav/home/Rat_cagescan/BAM/

Alignments uploaded on 30-Mar-2011 and 01-Apr-2011 was prepared by Charles, but the standard pipeline should produce identical or very similar results.

To retrieve properly paired 5′ ends: samtools view -b -f 0x0042 [input] > [output]

Clustering (provisional)

OSC tables for Level 1 promoters: https://fantom5-collaboration.gsc.riken.jp/webdav/home/Rat_cagescan/L1/

BED12 formatted clusters of (properly) paired-end reads : https://fantom5-collaboration.gsc.riken.jp/webdav/home/Rat_cagescan/CLUSTER/

HeliScopeCAGE

The RNA sample 10009-101B8 , <quote>Universal RNA - Rat Normal Tissues</quote>, was also used to prepare the HeliScopeCAGE library CNhs10614, available in BAM or BED formats.