Quality of Mapping Talk

Mapping quality control

Some issues are only detectable in the context of the genome:

• Duplicate reads
• Fragment size distribution
• Gene coverage
• Completeness of data

Duplicate reads

• Only detectable with paired end reads

Duplicate reads 2

• Duplicates can be PCR artefacts
• Duplicates can be real, from highly expressed transcripts
• For RNA-seq, removing duplicates is still being debated
• We don’t remove them, but it’s important to:

- assess the duplicate rate

- determine whether the duplicate rate can be explained by a few highly expressed genes

Fragment size distribution

* Should correspond with fragment size selected during library preparation Take into account that reads can span introns when calculating fragment size

Gene coverage

* Read coverage of the gene should be uniform Less coverage at the ends is expected because of degradation of the RNA There should be no 5' nor 3' bias

Completeness of data

* From a saturated RNASeq dataset, all known splice junctions should be rediscovered. Check saturation by resampling resampling 5%,10%,..,100% of alignments, detect splice junctions from each subset and compare them to reference gene models