Exome sequencing

[arvi8689]

Hey I am trying to sequence the melanoma exome to find somatic mutation involved in the melanoma. I am very new to this technique and I have some questions regarding this:

1)Generally how many exomes from different melanoma samples need to be done to get some satisfactory results?
2)What coverage is essential for such a study? Initially I was thinking of around 30X but some papers have used nearly 100X. Will I be okay with 30X coverage of the exome?

Your help is well appreciated.

[samesense]

I'm also doing melanoma exome analysis.

1) I think this depends on what you are looking for, where your tumors are located, and where you plan to publish. If you want CNV patterns, I'm seeing those with ~30 tumor/normal pairs. For novel somatic SNVs common across the tumors, you will need many more pairs, as the only common SNVs I'm seeing with 30 have already been reported. For a pathway analysis, you might be able to get away with 10 pairs, but tumor location matters. Tumors with sun exposure can have 100s of somatic mutations, while mucosal melanomas can have less than 10, leaving you with few mutations to map to pathways. Also, if you want to publish in Science or Nature, you'll have to go big, say over 100 pairs if you want to compete.
2) We have good results with an average of 65x coverage across the tumors using Illumina GAx, but there are plans to go higher.

[arvi8689]

Hey

Thanks for the reply. I am looking for somatic mutations and not CNVs. One adv with my melanomas are that they r induced melanomas in model organism and they have not been exposed to UV. So I will not be getting any of the somatic mutations due to UV. You think in such a case I can go ahead with 10 samples.
65x is the coverage you are using to look at CNVs. Will it be enough for looking at somatic mutations too?

[BetterPrimate]

1)Generally how many exomes from different melanoma samples need to be done to get some satisfactory results?
2)What coverage is essential for such a study? Initially I was thinking of around 30X but some papers have used nearly 100X. Will I be okay with 30X coverage of the exome?
Your help is well appreciated.

You might find that addressing more specific questions will be more rewarding to you.

Can you define "satisfactory results"? Head on over to Pubmed and look at what others have done for some ideas e.g. the paper by Wei et. al PMC3161250

Access and funding constraints probably will limit you to choosing from a few technology options e.g. you might be choosing between running a quarter plate vs half plate per sample on a SOLID. So it would be more fruitful to directly compare the options available to you.

Coverage is a very indirect measure. The key question you probably want to ask is "what is the false negative rate?" for each option you have available. Put another way, what percentage of bases will have insufficient depth to make a variant call. If a causative variant is among those bases, you'll miss it. You can make your own decision on what is acceptable but also do your lit. search and be guided by what others have done.

[ulz_peter]

totally agree with BetterPrimate, if 30x fold coverage would mean 30 reads on every base it would be ok, but chances are very high that you miss bases, because there are no reads on that site.
We do an approximate of 100x which yields good results for most bases.

HOWEVER: If you are searching for somatic mutations you should consider some other aspects: are you sure your desired mutation will be present in 50% of all the alleles sequenced? Do you want to detect mutations that occur in some cells only? What is the possible level of contamination from non-tumor cells (immunological cells)?