PDA

View Full Version : Best company for testing



tahir0010
04-25-2017, 10:46 PM
I did ancestry dna, but I did not like the test, and the results were very wrong for what we know to be our background is there any company that you would recommend is better then ancestry? Is ftdna pretty accurate.

Osiris
04-25-2017, 10:53 PM
Is your goal is the ethnicity percentages or to find genetic cousins to assist with family reasearch?

sktibo
04-25-2017, 10:57 PM
IMO ftDNA is the worst of the major companies, and 23andme is the best for autosomal ethnicity estimate.

A Norfolk L-M20
04-25-2017, 11:07 PM
In my opinion there isn't a good autosomal DNA test for ancestry - yet! None of them are accurate below regional, or sometimes even Continental level, although some appear slightly better for identifying a few specific populations.

Best option on autosomal is to take multiple tests, use admixture calculators, and also to look at any matches. Even then, never take the results as an indisputable truth.

mwauthy
10-27-2017, 12:56 PM
In my opinion there isn't a good autosomal DNA test for ancestry - yet! None of them are accurate below regional, or sometimes even Continental level, although some appear slightly better for identifying a few specific populations.

Best option on autosomal is to take multiple tests, use admixture calculators, and also to look at any matches. Even then, never take the results as an indisputable truth.

I noticed on websites that use the Joe Pickrell algorithm like DNA Land, Geneplaza, and Gencove I get high percentages of Southwestern European and North Slavic, Eastern European, and Ashkenazi. Yet on most of the commercial testing ethnicity estimates I donít.

Then it occurred to me that there does not seem to be any reference samples from Western or Central continental Europe. The Northwestern category seems to just involve the British Isles and Scandinavia. So if you have ancestors from Western or Central Europe your ethnicity estimate is going to show larger percentages from neighboring regions to substitute a type of false equilibrium for this discrepancy.

A lot of these companies including the commercial ones are misleading their customers about their true ancestry because they are assigning neighboring regions to make up for a lack of reference samples. Until these companies make the effort to get many more reference samples and to split up their current regions into smaller regions these ethnicity estimates are always going to be pretty incomplete and often misleading.

C J Wyatt III
10-27-2017, 03:30 PM
IMO ftDNA is the worst of the major companies, and 23andme is the best for autosomal ethnicity estimate.

Some may do admixture better than others, but if you are interested in matching to find relatives especially using GEDmatch, FTDNA Family Finder is the only one that is still producing what I consider a "standard" kit. The others are selling reduced SNP's (ones used for genealogy) kits.

I would not go anywhere near 23andMe's new V5 chipset kits that actually makes up results (imputation).

Jack Wyatt

MacUalraig
10-27-2017, 04:01 PM
Some may do admixture better than others, but if you are interested in matching to find relatives especially using GEDmatch, FTDNA Family Finder is the only one that is still producing what I consider a "standard" kit. The others are selling reduced SNP's (ones used for genealogy) kits.

I would not go anywhere near 23andMe's new V5 chipset kits that actually makes up results (imputation).

Jack Wyatt

No, the 23andMe chip contains tagSNPs that studies have proven are powerful predictors for other SNPs. The chips themselves don't impute so the calls in the raw data are all real genotyped calls.

C J Wyatt III
10-27-2017, 04:29 PM
No, the 23andMe chip contains tagSNPs that studies have proven are powerful predictors for other SNPs. The chips themselves don't impute so the calls in the raw data are all real genotyped calls.

Just a lot fewer SNP's than we are used to. OK, the imputation is being done outside of the chip.

The trouble with imputation is the same with admixture analysis. The experts are missing something really big. It is the reason why we have so many matches, but only can see how we are related with just a handful.

I'll take real SNP's, please, for my matching.

Jack Wyatt

Donwulff
10-28-2017, 08:54 PM
Lifestyle health and research with third party services, single test does all, may contribute to commercial research, , only recently available outside US/UK, possibly non-CLIA certifiied lab: AncestryDNA
Validated, clear and actionable health information for extra payment, advanced ancestry analysis used in peer-reviewed publications, rough maternal/paternal haplogroups, future-proof multi-ethnicity raw data, contribute to research: 23andMe
Compatibility with most other autosomal results with autosomal transfer, ability to match by autosomal and maternal/paternal line if tested, genealogy community projects: FTDNA
Up-and-coming, probably best family tree database and service with subscription, autosomal transfer, most same as FTDNA without maternal/paternal options: MyHeritage
Free with questionnaires in the USA, health and research with third party utils, genealogy matching with GedMatch genesis: Genes for Good
Something something, this seemed UK only so I haven't really looked at it before, but now they too are accepting imports: LivingDNA
Serious genealogy, adoptees: 23andMe Ancestry Only and AncestryDNA, autosomal transfer to FTDNA, MyHeritage, LivingDNA, GedMatch

According to Debbie Kennett, a renowned genetic genealocist https://cruwys.blogspot.com/2017/08/23andme-launch-new-v5-chip-and-revise.html the Illumina OmniExpress used by FTDNA is being phased out in favor of GSA, the new chip currently used by 23andMe and Living DNA. Be it as it may, the GSA has better coverage of multi-ethnic (Caucasian) SNP's, so it would give better matching results for multi-ethnic matches, so GSA will be favored test chip going forward. FTDNA is already handling a large amount of autosomal transfers from services with other chips, so they aren't free of the SNP confusion even now. In principle, only "matches" that represent non-genealogical background-sharing within population should be affected, but it does depend a lot on particular matching implementation.

N.B. If it matters to you, read the fine print on the Terms of Service carefully. There's lots of misinformation on this. Most services reserve the right to use non-identifiable genetic information in whichever way they want, as well as the right to change the Terms of Service any time. The kind of research that's needed to bring new drugs on market requires strict informed consent, and genetic data without associated survey data is pretty much useless for research, or they could just analyse biological waste and save the acquisition costs.