Repeat masking

[Amy S]

Hi everyone,

We have whole genome data from a number of tumour-normal pairs, and are looking at our score distributions for somatic changes. We find that some of the highest scores look like errors due to read pile up in repetitive regions when viewed in IGV. Thus we are beginning to think about masking repetitive parts of the genome using repeat masker or something similar.

Does anyone have any tips on how to go about this? I've had a brief look, and to me it seems I could mask up to 50% of the genome using the least stringent criteria. Does anyone have any experience to get me started? Which programs/repeat libraries are the best to use.

Thank you for your help,
Amy.

[jochensupper]

For homology based repeat masking I would use RepeatMasker (www.repeatmasker.org/) together with Repbase (www.girinst.org/repbase/). Its easy to use and usually provides very good results.

If you want to uncover de novo repeats there are many programs you could choose from, e.g., LTRfinder, PILER, RepeatScout.

[mattanswers]

When you do your alignment, you could align for only unique sequences (sequences that match to only 1 place on the genome). This can be done in bowtie by inserting -m 1.

Also, have a look at The Uniqueome: A mappability resource for short-tag sequencing,http://bioinformatics.oxfordjournals...s.btq640.short and here for a turorial:http://grimmond.imb.uq.edu.au/unique...ary_File_2.pdf