whole genome bisulfite sequencing in mouse: how many reads required

[gandalf886]

Hi,

I am going to do some WGBS in mouse, before I start, could you give an idea how many reads I need in order to have a good CpG coverage? or how many lanes of hiseq2000 per sample should I do?

Thanks

[nucacidhunter]

I would recommend looking at “NIH RoadMap Epigenomics Mapping Consortium” website: http://www.roadmapepigenomics.org/

As a ball point figure they recommend minimum of two biological replicates/treatment with combined 30x coverage (15x/strand). Most of available kits will result in fairly short insert (100-150 bp) bisulphite converted libraries suitable for 2x50 cycle sequencing. Assuming a 3 Gb mouse genome and 180 M read/lane output, each treatment will need approximately 5 lanes of HiSeq sequencing to obtain minimum recommended coverage.Taking into account mapping efficiency of 60-70% will increase lane number by 1.5-2 totalling 6.5-7 lane/treatment. So, for control and one treatment minimum of 13 lanes sequencing is required.

[dpryan]

It depends on what you want to do with the data. I have a dataset where I used a single flow-cell with one sample per lane. That'd be about the minimum I could ever recommend. Your mapping efficiency should be upwards of 80% with mouse data, though you'll end up using a smaller portion of that if you QC things properly. I ended up with ~6x CpG coverage per-sample (in total >30x).

Depending on what you're doing, you might consider targeted bisulfite sequencing. There's now a mouse kit to target promoters and such (note, this isn't RRBS), so that might be a nice alternative. The costs and analysis can both become a bit more appealing then.

[gandalf886]

thanks guys

I would recommend looking at “NIH RoadMap Epigenomics Mapping Consortium” website: http://www.roadmapepigenomics.org/

As a ball point figure they recommend minimum of two biological replicates/treatment with combined 30x coverage (15x/strand). Most of available kits will result in fairly short insert (100-150 bp) bisulphite converted libraries suitable for 2x50 cycle sequencing. Assuming a 3 Gb mouse genome and 180 M read/lane output, each treatment will need approximately 5 lanes of HiSeq sequencing to obtain minimum recommended coverage.Taking into account mapping efficiency of 60-70% will increase lane number by 1.5-2 totalling 6.5-7 lane/treatment. So, for control and one treatment minimum of 13 lanes sequencing is required.

It depends on what you want to do with the data. I have a dataset where I used a single flow-cell with one sample per lane. That'd be about the minimum I could ever recommend. Your mapping efficiency should be upwards of 80% with mouse data, though you'll end up using a smaller portion of that if you QC things properly. I ended up with ~6x CpG coverage per-sample (in total >30x).

Depending on what you're doing, you might consider targeted bisulfite sequencing. There's now a mouse kit to target promoters and such (note, this isn't RRBS), so that might be a nice alternative. The costs and analysis can both become a bit more appealing then.