I would like to detect exon duplication and deletion from NGS with panel amplicon strategy. For example, I have all exons of APC sequenced with an average depth > 100. ( exon 17 is covered by 34 amplicons ) How can I detect deletion of exon 3,4 or 5 ? is there any tool or method ?
My idea is:
- Split the exon region into bins ( or amplicon region )
- Count how many read fall into each bin
- Normalize bin count using %GC Lowess
- use normal sample data to create a model ( machin learning ... )
- Compare unknown sample to my model and check for a significant difference