ENLIS decoding mitochondria related genes...lost!

[TrixieStix]

Anyone else notice differences between mutations reported by Promethease compared to Enlis? I have found instances where Enlos reports a mutation Promethase did not report and also where Enlis seems to be labeling a mutation with a different snp # than Promethease did. I am still trying to figure out exactly what is going on.

 

[pattismith]

the Genetic Medicine Specialist I saw at a large univeristy teaching hospital told me that I could assume my 23andme mutation information is correct. But that saod I do know that there are a few known flukes in 23andme's reading of specific genes. A few are reported on snpedia.com as likely being misreported/flipped or something like that.

Hi Trixie, do you have rare missense mutation on the MT-ND1 and MT-CYB genes, or on any other mito DNA genes?

 

[TrixieStix]

@pattismith No variants of any kind reported for MT-ND1 nor MT-CYB. But I do have 2 novel protein-coding missense heterozygous variants on MT-ND2...one of them has an allele frequency of 1.90% and the other 97.97%

I do have a "rare" (less than 1% frequency) DNA variant for MT-ND5

What is the difference between a "DNA variant"and a "Protein-coding variant" ??

 

[pattismith]

Anyone else notice differences between mutations reported by Promethease compared to Enlis? I have found instances where Enlos reports a mutation Promethase did not report and also where Enlis seems to be labeling a mutation with a different snp # than Promethease did. I am still trying to figure out exactly what is going on.

I ran my data on Promethease as well, and the mutation on the nucleotide 15 198 on the MT-CYB gene didn't show up.
idem for the three mutations found with ENLIS on the MT-ND1 gene, they don't show up on Promethease...:thumbdown:

 

[Valentijn]

What is the difference between a "DNA variant"and a "Protein-coding variant" ??

A DNA variant can be anywhere on the genome, including far away from genes. A protein-coding variant will be a mutation in the exome, probably resulting in a missense mutation, nonsense mutation, or other substantial change to the protein created by the gene. There's a lot of gaps between genes on chromosomal DNA (1-22), but mitochondrial DNA is almost entirely protein-coding.

Anyone else notice differences between mutations reported by Promethease compared to Enlis?

Promethease seems to rely on different data, primarily published research. If there's no research showing how common a SNP is, it's possible that they aren't listing it. And a lot of suspicious SNPs haven't been directly researched at all yet. Enlis seems to be flagging rare mutations quite well from what I've seen, as well as labeling them accurately.

 

[Valentijn]

Another issue is that some databases will only include SNPs ... which technically have a minimum known prevalence rate in humanity. The number usually thrown around is 1%, but even dbSNP goes lower than that. But a lot of mutations won't be listed at all because they're virtually unknown to occur, except in extremely rare cases.

 

[TrixieStix]

A DNA variant can be anywhere on the genome, including far away from genes. A protein-coding variant will be a mutation in the exome, probably resulting in a missense mutation, nonsense mutation, or other substantial change to the protein created by the gene. There's a lot of gaps between genes on chromosomal DNA (1-22), but mitochondrial DNA is almost entirely protein-coding.

Thanks for the explanation. On the line in the gene report where it lists the "DNA variation" it has a box labled "Protein-coding synonymous". Any idea what that means?

 

[Valentijn]

Thanks for the explanation. On the line in the gene report where it lists the "DNA variation" it has a box labled "Protein-coding synonymous". Any idea what that means?

It means it's in a protein-coding section, but the change in the DNA doesn't result in a change in the protein. This happens because usually 2 or 4 different codons (combinations of three consecutive alleles) all create the same amino acid. So if CCC changes to CCG in a protein-coding section, it won't matter because both of them create proline. Hence there's no change, despite there technically being a mutation, and the amino acid (and protein) remains the same (synonymous).

 

[TrixieStix]

It means it's in a protein-coding section, but the change in the DNA doesn't result in a change in the protein. This happens because usually 2 or 4 different codons (combinations of three consecutive alleles) all create the same amino acid. So if CCC changes to CCG in a protein-coding section, it won't matter because both of them create proline. Hence there's no change, despite there technically being a mutation, and the amino acid (and protein) remains the same (synonymous).

Does this mean DNA variants don't cause disease/aren't pathogenic? Are they something that we don't need to pay attention to?

 

[Valentijn]

Does this mean DNA variants don't cause disease/aren't pathogenic? Are they something that we don't need to pay attention to?

It's possible for them to cause problems, but it's very rare for them to do so. In any event, there's not much data regarding which ones cause problems. So focusing on coding variants is a lot more efficient and productive for now.

 

[TrixieStix]

I have just found an instance where Promethease reported an SNP it labels as "bad" (it is a known risk factor for ARMD) but Enlis did not report this snp at all. The snp is rs800292 (it is on the CFH gene...complement factor H) Any idea why this would be?

I am noticing that Enlis is flagging some SNP's as a "risk factor" or "pathogenic" that on Promethease are listed as "not set" in terms of whether they are good/bad.

Also why does Enlis list a variant as "Pathogenic" or "Risk Factor" but then when you go to the detailed screen for that specific gene the variant says "no" under "predicted deleterious" ?? are they 2 different things?

Another thing... am I correct in assuming Enlis is always reporting the "Allele Frequency" as if it was heterozygous (1 variant) even when you are homozygous (1 variants)? Whereas with Promethease they seem to be reporting the "frequency" for variants based on whether you are heterozygous or homozygous for that variant.

 

[Valentijn]

I have just found an instance where Promethease reported an SNP it labels as "bad" (it is a known risk factor for ARMD) but Enlis did not report this snp at all. The snp is rs800292 (it is on the CFH gene...complement factor H) Any idea why this would be?

It's a very common missense mutation, and at most is associated with an increased risk of macular degeneration. But it doesn't cause disease on its own. So Enlis is probably filtering out that sort of stuff, because otherwise it is incredibly spammy.

also why does Enlis list a variant as "Pathogenic" but then when you go to the detailed screen for that gene the variant says "no" under "predicted deleterious" ?? are they 2 different things?

Pathogenic is a flag a SNP gets from nih.gov, based on assessing submissions that describe that SNP as benign, pathogenic, etc. It's not always accurate or up-to-date. Pathogenic flags are usually a better indication than predictions, because they are based on research. But it comes down to the quality of that research. Pathogenic doesn't always mean disease-causing either, in those databases. Eg, a mutation that causes red hair is labeled pathogenic.

Predictions are generally based on protein modeling software, and data of other nearby mutations, etc. Eg, @pattismith's "unknown" mitochondrial mutations discussed above might be (primitively) predicted as pathogenic due to being on the same codon as pathogenic mutations, and resulting in the creation of a substantially different amino acid. Predictions are educated guesses, but they can be extremely sophisticated. They're also very useful to tentatively fill in the huge gaps in genetic research. BLOSUM62 and other BLOSUM matrixes are very basic ways to make a prediction.

 

[TrixieStix]

@Valentijn I "edited" my last post and added a few more remarks/observations/questions

 

[Valentijn]

Also why does Enlis list a variant as "Pathogenic" or "Risk Factor" but then when you go to the detailed screen for that specific gene the variant says "no" under "predicted deleterious" ?? are they 2 different things?

I think I mostly answered this above, but basically it's different types of data from two different sources. I don't think it's possible to say which is more likely to be right without reading the pathogenic/risk factor/etc research.

Another thing... am I correct in assuming Enlis is always reporting the "Allele Frequency" as if it was heterozygous (1 variant) even when you are homozygous (1 variants)? Whereas with Promethease they seem to be reporting the "frequency" for variants based on whether you are heterozygous or homozygous for that variant.

Allele frequency is the normal way to calculate prevalence for a SNP. And that is referring to how often the allele appears in a sample of people. So if you have 50 people, the gene is on a chromosome where everyone has two alleles, there are 100 alleles being looked at in the study. If the "C" allele 5 times, the allele frequency is 5% (5/100).

Allele frequency doesn't tell you how common it is to be homozygous versus heterozygous (genotypes). It can be calculated, but calculations may be misleading in some situations. For example, a SNP might be deadly, but only if homozygous. So 5% of people might be heterezogyous for it, but 0% homozygous. Hence if someone is listing homozygous and heterozygous rates, they're really just making a calculation which is often not very precise. dbSNP gives some actual genotype rates with various cohorts, typically.

 

[TrixieStix]

@Valentijn I was just making an assumption about the "allele frequency" based on seeing that the "frequency" %'s given by the 2 different platforms vary greatly from one another. Hmmm?

 

[Valentijn]

@Valentijn I was just making an assumption about the "allele frequency" based on seeing that the "frequency" %'s given by the 2 different platforms vary greatly from one another. Hmmm?

Promethease might be using different data from SNPedia. If so, it's a lot less accurate than data used from dbSNP.

 

[BeautifulDay]

Another way of sorting data is to look for rare variations on Enlis (that are not clinically significant/pathogenic - so it wasn't yet pulled out by the above method) that are also on the MitoD database. This is extremely time consuming, and often provides false leads due to either 23andme having an error in calling the snp, or it being reversed for the rare allele, or Enlis calling it rare when it's not. I have no idea which it is or some combination of all of the above.

Here is an example. I filtered my 23andme results to look for rare mutations (2% or less of the population has it) on Enlis where I was homozygous for having it. So a very rare combination. Then I ran those very rare mutations where I was homozygous to see if they occurred on any of the genes that were in the MitoD database.

I came up with 11 of them. That is running my 23andme data through Enlis came up with 11 homozygous (x2) rare mutations that were also on the MitoD database. You would think if it was this rare, my mom would be heterozygous (x1) for these, and my father would be heterozygous for these, and my husband wouldn't have any of these. Nope, for the most part everyone in our family was x2 (homozygous) or almost everyone was x2 for these supposedly rare mutations. Therefore, these get thrown out.

The 11 snps I found useless where it said I was homozygous rare were:
rs189362, rs2126115, rs5927969, rs5928196,
rs2806233, rs13066381, rs17064653, rs11749784, rs439954, rs244411, rs8082645

Therefore, using 23andme data this way didn't help in my search.

 

[TrixieStix]

Another way of sorting data is to look for rare variations on Enlis (that are not clinically significant/pathogenic - so it wasn't yet pulled out by the above method) that are also on the MitoD database. This is extremely time consuming, and often provides false leads due to either 23andme having an error in calling the snp, or it being reversed for the rare allele, or Enlis calling it rare when it's not. I have no idea which it is or some combination of all of the above.

Here is an example. I filtered my 23andme results to look for rare mutations (2% or less of the population has it) on Enlis where I was homozygous for having it. So a very rare combination. Then I ran those very rare mutations where I was homozygous to see if they occurred on any of the genes that were in the MitoD database.

I came up with 11 of them. That is running my 23andme data through Enlis came up with 11 homozygous (x2) rare mutations that were also on the MitoD database. You would think if it was this rare, my mom would be heterozygous (x1) for these, and my father would be heterozygous for these, and my husband wouldn't have any of these. Nope, for the most part everyone in our family was x2 (homozygous) or almost everyone was x2 for these supposedly rare mutations. Therefore, these get thrown out.

The 11 snps I found useless where it said I was homozygous rare were:
rs189362, rs2126115, rs5927969, rs5928196,
rs2806233, rs13066381, rs17064653, rs11749784, rs439954, rs244411, rs8082645

Therefore, using 23andme data this way didn't help in my search.

what variant did end up being the important one and how did you single it out of the crowd?

Also can you post a link to the MitoD database?

 

[pattismith]

I wonder if you may have also a missense mutation in the OMA1 gene, like the one I have?
It's still about mitochondria function

 

[TrixieStix]

A DNA variant can be anywhere on the genome, including far away from genes. A protein-coding variant will be a mutation in the exome, probably resulting in a missense mutation, nonsense mutation, or other substantial change to the protein created by the gene. There's a lot of gaps between genes on chromosomal DNA (1-22), but mitochondrial DNA is almost entirely protein-coding.


Promethease seems to rely on different data, primarily published research. If there's no research showing how common a SNP is, it's possible that they aren't listing it. And a lot of suspicious SNPs haven't been directly researched at all yet. Enlis seems to be flagging rare mutations quite well from what I've seen, as well as labeling them accurately.

I am still baffled as to what is going on in regards to my CFH gene (complement factor H). Promethease shows me as homozygous for rs800292, but Enlis does not report this snp at all ,and instead shows me as homozygous for rs1061170. ??? Both of these variants are well studied and are well known to confer elevated ARMD risk so each site would be expected to report these variants if I indeed had both.
Th
e only thing I can think is that one of the sites is reading the raw data incorrectly, but which one?