Whilst considering the answer to a recent question about polymorphism definition across databases it occurred to me that I do not know how genotyping techniques can determine a polymorphism or inherited variation from a somatic mutation. I do not know if it is possible to distinguish an infrequent variation from a somatic mutation when sequencing/genotyping sequencing an individual. The only distinction I have ever seen in variation repositories is that HGMD does not store somatic variations.
So to recap:
1)when genotyping, is it possible to distinguish an infrequent variation from a somatic mutation
2)are somatic variations handled differently by any database other than HGMD
(I have seen the related question on this website regarding polymorphism vs infrequent mutation).
What do you mean by the real problem behind this? I am about to be given a lot of data with somatic mutations in from a specifc disease tissue so I'm trying to get a feel for the area. If there is a 'real problem' I'd really like to know about it!
As you see from my answer, to separate recent/new mutations from frequency-defined polymorphisms can be very very tricky if you don't have proper controls. Check out some best practices in http://www.cmgs.org/BPGs/Best_Practice_Guidelines.htm
yes i can see that. was just wondering if you meant anything else. This question has been very helpful. I wish i could split the answer between you and David.
Your use of "somatic" in this context seems to be different from the standard definition (e.g. see this definition.
"Germline DNA" is typically read to mean "the DNA sequence you were born with". The term "somatic mutation" is usually read to imply that the germline DNA sequence was altered somehow after you were born; that's where the "somatic" part comes in. Part of the definition, in my mind, is that this mutation is not present in every cell of the body. Somatic mutations are typically identified by comparing the sequence of a particular pool of DNA-- let's say DNA extracted from a lung tumor-- to DNA thought to be representative of the germline DNA. For people with epithelial tumors, DNA extracted from blood is usually used to get the germline status. You can identify novel somatic mutations by comparing the target sequence to germline sequence. Somatic mutations are catalogued in databases like COSMIC.
This definition is distinct from the question of "what's the difference between a common polymorphism and a rare polymorphism?", since the context for that question is the DNA the person was born with. That question is usually a matter of commonly accepted definitions for "rare" vs. "common" and the population being used for comparisons, rather that being a difference in kind. Whatever the definition, I think these unusual variations are usually termed "rare variants" rather than "somatic mutations".
I'm not a wet-lab biologist, but the basic method to make LBCL would start by extracting a whole blood sample. This has several cell types in it, including red blood cells and white blood cells (part of the immune system). Extract out the B cells (part of the immune system) by centerfuge, infect with Epstein-Barr virus to immortalize the cells, and culture the cells. Then you can extract DNA from those cells using standard protocols. Note that LBCL are, by definition, not normal cells and alterations to the DNA and gene expression occur during immortalization and culture.
David - I do indeed mean the exact definition that you point out and I totally agree that a somatic mutation will only be present in a specifc tissue and not all tissues. The problem is probably because i am not a biologist and don't fully know how you get an individuals's DNA to create the cell lines that are sequenced. What tissue do you use? If you get DNA from a germline cell won't the individual only be haploid? If you get any other type of tisue isn't there a chance of somatic mutations being present?
Also i'm aware that a rare variant is not a somatic mutation. I pointed out that I had read the other question as it was semi related and many respondents tend to point out similar questions and I thought i would save them the trouble! :)
Normally (as I mentioned) the individual's germline DNA is identified by extracting DNA from a blood sample. Blood is easy to get. One wouldn't normally derive a cell line from blood (e.g. a lymphoblastoid cell line, as in the CEPH samples) unless you wanted a renewable supply of sample. There is a chance that somatic mutations would have occurred in the blood cells you sample, but the assumption is that outside of a clonal tumor, these mutations would be very rare and confined to individual cells compared to the total DNA population. We're assuming the subject doesn't have a blood cancer.
I was actually thinking of the lypmphoblastic cell lines when i asked this question. Where do you get teh DNA to create those from? Thanks for clearing up the issue with the blood sample. So really its a numbers game and the blood cells could have somatic mutations but they are rare and confined to individual cells. Makes sense. Is there a figure for the rate of somatic mutations? I know the rate of base changes in ~10^9 per base per year. Is that the same thing as the rate of somatic mutation?
1) It's possible to genotype someone and detect a new mutation. By definition, a polymorphism (SNP in HapMap sense for example) must be segregating in the population at a certain frequency, a mutation don't. Unfortunately, a lot of mutations are quite recurrent in the human population (e. g. in the RET gene) and easily confounded with SNPs. By frequency alone, it's impossible to distinguish a somatic recurrent mutation from a SNP. To complicate matters, the effective number of alleles (genic equivalent of effective population size) per locus varies wildly along the genome. The very definition of rare/common is much more subtle than it seems.
So, frequency is not enough to tell the difference, you'll just need to gather some genealogical data. If neither the mother/father/original tissue has it, it must be a new mutation. Given that mutation is a difference, a change in character, you must have a reference. Genealogy is the best one.
2) By now, I'm not a 100% sure , but I think that each database has it's own criteria to handle somatic mutations. This means that interoperability and congruence isn't guarateed. I'm trying to gather definitions of polymorphism/mutation from major DBs. Only those from NCBI stated it explicitly!!!
so when there is no frequency data available, we don't know if the SNPs in dbSNP are variations or somatic mutations in the absence of genealogy. You will see from my comment to David's answer that I don't know whether biologists are able to isolate germline DNA when they carry out genotyping, because DNA from any other tissue could possibly have somatic mutations i think?
In other words, if you could be sure you had germline DNA you would know you have a variation but if you had any other tissue it could contain somatic mutations
That's why I'm very worried with definitions and such. E.g. The typical control tissue in oncology is blood, a tissue full of crazy stuff. When one genotype using Sanger, things are less complicated. But, with NGS you can go into trouble very very easily . . .
that's why the variant filtering is so critical, specially if you aim to detect mutations with relevant disease association. unfortunately we are still far from getting a final result list just from a completely automated machine.
is variant filtering a 'special' process i should know about or does it refer to the general notion of looking through your variants and throwing out the ones you don't want.
Hum ... This is a very recurrent question. I think most of us simply don't get the real problem behind this ...
What do you mean by the real problem behind this? I am about to be given a lot of data with somatic mutations in from a specifc disease tissue so I'm trying to get a feel for the area. If there is a 'real problem' I'd really like to know about it!
As you see from my answer, to separate recent/new mutations from frequency-defined polymorphisms can be very very tricky if you don't have proper controls. Check out some best practices in http://www.cmgs.org/BPGs/Best_Practice_Guidelines.htm
yes i can see that. was just wondering if you meant anything else. This question has been very helpful. I wish i could split the answer between you and David.