Sea
Senior Member
- Messages
- 1,286
- Location
- NSW Australia
@Firestormm a few thoughts in response to some of the issues you raised
The relationship with Google - The founder of 23andme was until recently the partner (personal) of the founder of Google. They saw the business potential for lots of profit in this area in the future, which is why this venture is in part sponsored by Google.
$99 - People keep saying some version of this "what do you expect for $99 dollars?" The true value of the test is way above this. It used to sell for $299 + an annual paid subscription to keep access to results before they got a wealthy sponsor involved but even that is below cost. Their subsidisation of the cost is a business investment. The true value for the company is in the data they will have when they have a million customers who give them permission to use their genetic info. Researchers will be lining up to get access to it. I don't have a problem with entrepreneurs making money from my data as long as there is some reciprocal benefit for me (knowledge not money)
As a comparison, they test 960 000 snps for $99. Most single gene tests for disease by other companies are in the vicinity of several hundreds to several thousands of dollars. Some of those are gene sequencing, some are testing snps. Yasko, who is mentioned extensively on this forum, tests several dozen snps for around $400. A test for Haemochromatosis in Australia is $70 and tests exactly the same 3 snps as what is included in 23andme (without the 959 997 others
)
"Veneer of science" - I don't think that's a fair assessment. The science is in it's infancy and needs a very good understanding of how to use statistics and of the potential pitfalls before any conclusions can be drawn. There is less information available for rare snps than there is for common ones. If a snp is common it takes a huge number of people to detemine any disease correlation. If a snp is rare it has usually been less studied and there are fewer people available to study it in. The human genome project has a database of 1000 human full genome sequences. That is not enough to study disease correlations in mutations that occur in less than 1% of people unless there is a very direct cause and effect like there is in Cystic Fibrosis for example.
"Judging by the threads on this forum" - Much of what is discussed on here uses the raw data from 23andme but not the interpretations that come from 23andme. I think it's fair to say that some here have a good grasp of science and know how to follow research appropriately and others don't have a clue and reach some far fetched conclusions from what they learn. There is much more information available about the snps tested than what 23andme actually report on but it takes a lot of research to come to reliable conclusions.
23andme are providing recommendations and I think they have mismanaged how they have worked with the FDA on that. They definitely should not have made the advertising statements they have about health and individualised recommendations if they didn't want to raise the hackles of the FDA
I know many people misunderstand the risk calculations by 23andme. I think many who look at their results do not move beyond the overview to understand what the risk assessments are based on and what they really mean. 23andme does differentiate between established research, preliminary research, controversial research where different researchers have reached opposing conclusions and data gathered from 23andme customer surveys.
They note whether the research has been replicated or not. They provide links to studies in all their reports. They use symbols to denote the size of the data pool that the research is based on. They make it clear that their information is for educational and research purposes only and not to be used to diagnose or treat and that someone with concerns should take the information to their doctor and discuss it with them.
Personally I think there is much room for improvement in providing information to a not necessarily scientifically literate customer base, but I would hate to see this information being guarded and only available through a doctor.
As an example if I look only at my overview it tells me I have a reduced risk for Celiac disease.
My risk 0.13% Average risk 0.24%. Therefore my risk is 0.54% of average. It does not mean I don't have Celiac.
When I look at the more detailed report it tells me that they base their risk assessment on 4 established markers (some of mine are the riskier version and some are not). Research is linked to for each marker. It is made clear that their risk assessment for the HLA marker is based only on the most common risk factor (HLA DQ2.5 - carried by 30-40% of the general population but present in over 90% of people with Celiac) but that there are other HLA DQ risk versions which some people with Celiac have (HLA DQ2 & 8) I don't have DQ2.5
Many of the genetic testing companies that doctors refer people to for Celiac testing only test for DQ2.5 and people are told their risk has been ruled out if they don't have that. That is simply not true and mainstream testing is beginning to recognise that.
From further research with my raw data I discovered I most likely have the HLA DQ2.2 version which accounts for about 5% of people with Celiac. I have 2 out of the the 3 necessary markers, the 3rd is not covered by 23andme. It was enough for my doctor to more seriously consider more thorough testing though I have no family history.
I am really pleased with the information I can learn from my data. I come from a family with a complex medical history with little help from doctors and I have been able to make sense of a lot from my genetic data. As soon as it is reasonable to have my whole genome sequenced I will be doing that. I expect my children will benefit from it more than I will as the research catches up with the information that is there waiting to be discovered. Even now though there is much to be gained.
Sorry, that was more than a few thoughts
The relationship with Google - The founder of 23andme was until recently the partner (personal) of the founder of Google. They saw the business potential for lots of profit in this area in the future, which is why this venture is in part sponsored by Google.
$99 - People keep saying some version of this "what do you expect for $99 dollars?" The true value of the test is way above this. It used to sell for $299 + an annual paid subscription to keep access to results before they got a wealthy sponsor involved but even that is below cost. Their subsidisation of the cost is a business investment. The true value for the company is in the data they will have when they have a million customers who give them permission to use their genetic info. Researchers will be lining up to get access to it. I don't have a problem with entrepreneurs making money from my data as long as there is some reciprocal benefit for me (knowledge not money)
As a comparison, they test 960 000 snps for $99. Most single gene tests for disease by other companies are in the vicinity of several hundreds to several thousands of dollars. Some of those are gene sequencing, some are testing snps. Yasko, who is mentioned extensively on this forum, tests several dozen snps for around $400. A test for Haemochromatosis in Australia is $70 and tests exactly the same 3 snps as what is included in 23andme (without the 959 997 others
)"Veneer of science" - I don't think that's a fair assessment. The science is in it's infancy and needs a very good understanding of how to use statistics and of the potential pitfalls before any conclusions can be drawn. There is less information available for rare snps than there is for common ones. If a snp is common it takes a huge number of people to detemine any disease correlation. If a snp is rare it has usually been less studied and there are fewer people available to study it in. The human genome project has a database of 1000 human full genome sequences. That is not enough to study disease correlations in mutations that occur in less than 1% of people unless there is a very direct cause and effect like there is in Cystic Fibrosis for example.
"Judging by the threads on this forum" - Much of what is discussed on here uses the raw data from 23andme but not the interpretations that come from 23andme. I think it's fair to say that some here have a good grasp of science and know how to follow research appropriately and others don't have a clue and reach some far fetched conclusions from what they learn. There is much more information available about the snps tested than what 23andme actually report on but it takes a lot of research to come to reliable conclusions.
23andme are providing recommendations and I think they have mismanaged how they have worked with the FDA on that. They definitely should not have made the advertising statements they have about health and individualised recommendations if they didn't want to raise the hackles of the FDA
I know many people misunderstand the risk calculations by 23andme. I think many who look at their results do not move beyond the overview to understand what the risk assessments are based on and what they really mean. 23andme does differentiate between established research, preliminary research, controversial research where different researchers have reached opposing conclusions and data gathered from 23andme customer surveys.
They note whether the research has been replicated or not. They provide links to studies in all their reports. They use symbols to denote the size of the data pool that the research is based on. They make it clear that their information is for educational and research purposes only and not to be used to diagnose or treat and that someone with concerns should take the information to their doctor and discuss it with them.
Personally I think there is much room for improvement in providing information to a not necessarily scientifically literate customer base, but I would hate to see this information being guarded and only available through a doctor.
As an example if I look only at my overview it tells me I have a reduced risk for Celiac disease.
My risk 0.13% Average risk 0.24%. Therefore my risk is 0.54% of average. It does not mean I don't have Celiac.
When I look at the more detailed report it tells me that they base their risk assessment on 4 established markers (some of mine are the riskier version and some are not). Research is linked to for each marker. It is made clear that their risk assessment for the HLA marker is based only on the most common risk factor (HLA DQ2.5 - carried by 30-40% of the general population but present in over 90% of people with Celiac) but that there are other HLA DQ risk versions which some people with Celiac have (HLA DQ2 & 8) I don't have DQ2.5
Many of the genetic testing companies that doctors refer people to for Celiac testing only test for DQ2.5 and people are told their risk has been ruled out if they don't have that. That is simply not true and mainstream testing is beginning to recognise that.
From further research with my raw data I discovered I most likely have the HLA DQ2.2 version which accounts for about 5% of people with Celiac. I have 2 out of the the 3 necessary markers, the 3rd is not covered by 23andme. It was enough for my doctor to more seriously consider more thorough testing though I have no family history.
I am really pleased with the information I can learn from my data. I come from a family with a complex medical history with little help from doctors and I have been able to make sense of a lot from my genetic data. As soon as it is reasonable to have my whole genome sequenced I will be doing that. I expect my children will benefit from it more than I will as the research catches up with the information that is there waiting to be discovered. Even now though there is much to be gained.
Sorry, that was more than a few thoughts
