Difference Between SNP and Single Point Mutation in Cancer

Question

[Irmak]

I am working on breast cancer and try to detect somatic point mutation in FFPE samples. I am just confused about defining somatic point mutation and single nucleotide polymorphism in cancer tissue. Unfortunately I could not find any reliable source to eliminate the confusion. For instance,the same alteration classified as a polymorphism or mutation in different articles.

• For detection somatic point mutation, does it need to be use blood sample from same patients to detect ıf is it somatic or germline mutation? After comparison how can I be sure is it polymorphism or a mutation?
• If I say polymorphism what kind of material need to used? Blood or FFPE samples? What kind of control I need to use, just blood from healthy people or normal breast tissues?

Answer 1

[DG]

Don't worry too much about the terminology used in different articles. In my experience the terms are often used interchangeably and also somewhat incorrectly, usually depending on the background of the person writing the article. Typically researchers are often using the term Single Nucleotide Variant now for just this reason, because we may be variant calling in different contexts and calling it a SNP would be inappropriate when we are not talking about germline variants. The term mutation is also, technically, generic, although in the human context geneticists tend to want to reserve it for deleterious variants (coming from an evolution background myself I disagree but that is a separate issue).

Without matched normal tissue (usually blood, but not always) it can be difficult to confidently determine whether a particular variant is germ line or somatic, but it depends on what you are doing as to how important this is. Tumour only sequencing is possible, and typically we use a variety of filters to try and remove as many germline mutations as possible from the dataset. Usually this is done by filtering out all mutations seen in ExAc, 1000 Genopmes, EVS, etc. It won't get everything but it will remove quite a few.

But yes, if your study or use case requires a high-confidence set of somatic only variants you would probably want to sequence either from blood (if possible) or matched normal tissue from the same individual.

Answer 2

[Ibrahim Tanyalcin]

Dear Irmak,

If the variant exists in a subpopulation of healthy individuals, then you can call it single nucleotide polymorphism. A single nucleotide variant however does not tell you about the deleteriousness of the mutation. Even in the case of polymorphisms, certain combination of them (in different genes/loci) can have an adverse affect on the individual, so I would say it is safer to call SNV (single nucleotide variant) all together. I hope this helps,

Regards,

[DG]

In this case the OP is asking specifically about situations where we are sequencing tumors, so we are dealing with somatic variants. In this case it isn't about polymorphisms or deleteriousness, but about germline versus somatic variation.

[Ibrahim Tanyalcin]

Dear Dan,

There are SNPs that we still have no idea whether if they contribute to a disease phenotype as in GWAS studies. There are also cases of heterogeneity in a tumor sample where we detect multiple mutations but have no idea which one is the driver or even whether responsible for the malignancy. Sometimes in-vitro studies performed to replicate the effect of a particular mutation found in a tumor sample completely fails to show transforming behaviour. It does not matter in which context you use SNP and SNV, it is safer to use SNV.

SNV does not state anything about the origin(germline/somatic) or status(deleterious/benign) anyway. It just states it is a variant.

I hope this helps,

[DG]

I know, I pointed out to the OP that SNV was the more correct (and safe) term to use. Somatic mutations (or suspected saomatic mutations) in general shouldn't be called SNPs because they are not population level variants. My point was that the OP wasn't talking about the debate over say SNP versus mutation nomenclature as usually applied when dealing with Mendelian diseases, but specifically terminology confusion they were seeing in the cancer genomics literature

Answer 3

[cyril-cros]

Just another point about SNPs: they are defined as DNA sequence variations between individuals in a given population, and they have to be found in a significant fraction of your particular population (more than 1% typically).

The key point is that there will be more variants specific to the individual than just the SNPs, plus the somatic mutations specific to the tumor. SNPs are used as convenient shared markers for GWAS/evolutionary genetics.

I would absolutely use SNV and not SNP.

Answer 4

[poisonAlien]

As many have answered above, SNP generally tends to give impression that the observed variant is found in certain subset of the population as in dbSNP entries. However there are no strict guidelines on the usage of the term. Even TCGA uses Variant_Type as SNP for somatic single nucleotide changes in their MAF files. I have seen post docs in my lab asking why it is classified as SNP if its somatic. Using SNV instead seems to be fine for now.

[DG]

In the TCGA's case I believe it is actually just because that is how SNVs are typically annotated by most variant calling and annotation software, so that is how it is labelled in the VCF files.

Answer 5

[Irmak]

Thanks for answers. What about ıf I try to detect rs1800371 in FFPE tumor samples? What kind of control I need to use? Blood or normal breast tissue in FFPE from healty people?

[DG]

For simple detection purposes you don't need a control. You can identify whether the variant is present without it just by looking at the tumour. If you want to differentiate between it being a somatic mutation and germline then you need a matched control (or controls) from the same people whose tumour you are sequencing. Blood is usually the best bet unless you don't have it, and can't get it.

[DG]

And given that it looks like maybe that variant has been observed as germline in a small number of individuals from the ExAc datasets it would probably be a good idea to try and sort out whether the mutation was germline or somatically acquired. If you don't have blood but the FFPE block has surrounding normal tissue (as defined by a pathologist) it will do in a pinch. But blood is definitely preferable.