Announcement

Collapse
No announcement yet.

Anyone know what upcoming FamilyFinder changes are?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • nolnacsj
    replied
    Originally posted by mkdexter View Post
    When these sites can accept the new data then yes raw data should be re-uploaded. Not sure when they will be ready for the new data. They haven't seen the raw files yet which should be out any day now.
    Matt, thank you for your reply to my question.

    Judy

    Leave a comment:


  • Javelin
    replied
    Originally posted by JohnG View Post
    The previous discussion of corrections has implications to me. One is that the testing firms need to give users the data with a minimum of hidden assumptions and manipulation. That likely means providing more matches (maybe I want to know who I match in 200 2 cM segments, that might mean something; or maybe I only want to look at relations with 15 cM or more segments and I will take that as the only matching segment) and letting us filter them out. Another is that a good filter and management system is going to be complex in the long run.

    If the output of the matching run was available, and there were tools to manage it you might have lots of options.
    I agree, more tools are useful. In fact, Family Finder already has a centimorgan filter now (1+ cM, 3+ cM, 5+ cM, 10+ cM). deCODEme also has an excellent feature that allows you to compare your identical segments to pepole with whom you share, including sample populations, at various thresholds (I think it is 1 Mb, 3 Mb, 5 Mb, 10 Mb, 20 Mb). You can also visualize your data with gedmatch.com, an excellent offsite tool.

    My main problem with 23andme, in contrast, is that they do _not_ show all your matches at the higher thresholds, and they do _not_ let you filter your matches this way. In consequence, if you have certain ancestry, the preponderance of your "relatives" may come out of the same restricted gene pool from 500+ years ago, but are not closely related in any genealogical sense.

    Leave a comment:


  • JohnG
    replied
    New matches/corrections

    My first cousin once removed just sent in a sample. The pending date in her record says the batch will be done in early April, but she got an email saying it was in the new system and could be done next week. That would mean new matches coming out soon for some people.

    The previous discussion of corrections has implications to me. One is that the testing firms need to give users the data with a minimum of hidden assumptions and manipulation. That likely means providing more matches (maybe I want to know who I match in 200 2 cM segments, that might mean something; or maybe I only want to look at relations with 15 cM or more segments and I will take that as the only matching segment) and letting us filter them out. Another is that a good filter and management system is going to be complex in the long run.

    If the output of the matching run was available, and there were tools to manage it you might have lots of options.

    Leave a comment:


  • mkdexter
    replied
    Originally posted by nolnacsj View Post
    Matt, I understand you could not be more specific, and I appreciate what you do say in an effort to keep us informed as much as possible.

    OK, I have a question...how will the new platform effect the independent projects like Eurogenes, HIR, DODecad, and the GEDMatch tool? Will we need to upload our data again when we receive the Omni platform results (even if there are no new matches?).

    Judy
    When these sites can accept the new data then yes raw data should be re-uploaded. Not sure when they will be ready for the new data. They haven't seen the raw files yet which should be out any day now.
    Last edited by mkdexter; 11 February 2011, 01:52 AM.

    Leave a comment:


  • nolnacsj
    replied
    Originally posted by mkdexter View Post
    All the tests that are coming out next week are already on the new chip. Sorry I could say anything on this sooner. They've been working on this for a long time, hence the past FF results delays. They already have tests back with the new chip.
    Matt, I understand you could not be more specific, and I appreciate what you do say in an effort to keep us informed as much as possible.

    OK, I have a question...how will the new platform effect the independent projects like Eurogenes, HIR, DODecad, and the GEDMatch tool? Will we need to upload our data again when we receive the Omni platform results (even if there are no new matches?).

    Judy

    Leave a comment:


  • Javelin
    replied
    That's the crux of the issue. Currently 23andme applies an algorithm to all people with over a certain number of matching segments which artificially distances known close relatives. Unfortunately, the Relative Finder system also has a limit to sharers, and appears to favour many short segments over single longer ones (your cloud matches vs. tree matches).

    In practice, people with many matching segments may be entirely Ashkenazi, or entirely Mennonite, or entirely early colonial US, or 1/4 Ashkenazi 1/4 Finnish 1/2 Quebecois. Whatever the origin, it's important matches be treated as reciprocal. All other things being equal, any two people who share a 12 cM match share a pretty recent ancestor, regardless of their ethnic group and of the number of 6 cM segments you add to that. One 12 cM match is worth any number of 6 cM matches. That's something FTDNA gets -- 23andme, not.
    Last edited by Javelin; 10 February 2011, 06:47 PM. Reason: message truncated

    Leave a comment:


  • mkdexter
    replied
    In general terms correcting for Ashkenazi Jewish means adjusting the prediction so that a predicted cousinship range is shown as more distant that it was before the adjustment.

    Remember FTDNA does not adjust for this right now nor is there anything indicating they will in the future. They've just considered some ideas for research and for making predictions more accurate but have not committed to any special AJ adjusting method in the future.

    What companies are trying to compensate for are cases when most of the predictions are calculated too close by compounded segments or large sums of ancestral DNA. What it will eventually come down to is that every case is going to be different so there will need to be a method used to figure out which cases need which adjustment.
    Last edited by mkdexter; 10 February 2011, 06:13 PM.

    Leave a comment:


  • JohnG
    replied
    AJ and other corrections

    I wonder if someone could explain the proposed correction for Ashkenazi Jewish ancestry. I realize that in general Ashkenazim are related somewhat similar to 4th or 5th cousins. (I think that is true of any population that breeds within itself for geographic or cultural reasons). I seem to have some Ashkenazi ancestors myself and that makes the triangulation lists long. It also adds to the match lists.

    I have something similar with other ancestries. I have lots of early Massachusetts Bay ancestors and 8th cousins and 10th cousins match and we can find the link because of the great paper research over the centuries. I have something similar in colonial New Amsterdam, and in Germans from the Palatinate, and maybe in Scandinavia. For New Amsterdam there are good records and I found real cousins. Germany is tougher.

    I think in the long run we will all have too many matches to handle without a software management system. Whether that comes from FTDNA or elsewhere will be interesting to see. Is it that AJ and other close groups have more matches and run into the too many to handle limit earlier?

    Right now I look at the total cM and use its relation to the longest segment to categorize my matches into tree folk or cloud folk. The tree folk have some logical relation between total cM and longest that falls along the main spectrum expected as both reduce with more generations from the common ancestor. The cloud folk show up on a different line - they tend to have lots more total cM than fits the longest segment. I match them with a 8 cM longest segment and 40-80 total cM, for example. I like the cloud matches too, although it is unlikely I will build the step by step paper trail to connect many of them. They are sort of like haplogroups, telling me something more general about my ancestory. And sometimes they may get me to an area. I now have several matches that point to Lviv, Ukraine and I have an odd family name that occurs there.

    So, is the AJ correction just to reduce the matches that must be handled, or is there more to it? I can imagine general filters that can be set on total cM or total to longest ratios.

    And I wonder about those of us of varied ancestry. If the filter is not something we apply ourselves we may lose matches on on line while filtering out cloud connections on another.

    Leave a comment:


  • Kathleen Carrow
    replied
    As of a minute ago, 6 of the 10 folks with kits at the LowerDelmarva group have joined..

    Leave a comment:


  • Javelin
    replied
    ^ Agreed. FTDNA is focused on genealogy and is receptive to customers. 23andme had promise but is showing signs of getting out of genealogy to focus more or less totally on the medical side (that's my take on the "one version" and the monthly subscription, anyway).

    Leave a comment:


  • Kasandra
    replied
    Originally posted by Kathleen Carrow View Post
    Hmmm..that is not actually what I heard them say.. My feeling is that they are trying to resolve the Ashkenazi correction which surely FTDNA will also have to do as their database grows...someone told them they could NOT remove AF as the new subscriptions are for that but I believe they are for Relative Finder and the cousin matches.Ancestry Finder was always experimental AND had a large group of Beta testers.

    A lot of the angry talk at 23andme would NOT be tolerated here..Those of us here for years know that..

    That aside..I commend FTDNA for taking this large step forward..Those of us who have tested with the V3 chip would LOVE to be able to compare..I personally was assured by the "big guy" 3 times I would be able to upload and would be super thrilled to do that..and waited month after month..
    Congrats FTDNA..
    I didn't say they threatened to cut RF, I said AF. And the whole system is so freaking clunky and poorly assembled it's shocking. AF shouldn't be 'beta' it should be all integrated into RF so you don't have to go looking to see under the skirts so to speak. They have those other add ons because they didn't bother to upgrade it to one, easy-to-use, program. I agree that some folks may be extremely angry, but why is that? It's because they are frustrated in a way that no one here is. I'm not saying either service is perfect, the additional SNPs and larger database at 23 are valuable. But at the end of the day I have 12 more matches there than here. And a bunch of anonymous numbers that do me no good at all. I don't regret testing there because I value those matches, their customer service reps are usually, but not always nice. But their priorities suck rocks. They don't even try to balance the priorities and they don't understand the motivations of the genealogy side of things. They were CALLUS in their replies to me as if they had no idea why it would be upsetting not to be able to contact a second cousin or to lose a third cousin or have my 5, 6, and 7th closest matches revert to distant. They just said, oh well someone else will match you later. I felt the same as if someone said, "we're going to shoot your puppy but don't worry, we'll get you another one eventually."

    I believe that FTDNA, if they have to do something about AJ matches, will give the customer options to apply the filter or not because they can't afford to torque off their genealogy folks.

    Leave a comment:


  • Kathleen Carrow
    replied
    Originally posted by mkdexter View Post
    I posted once that FTDNA has some ideas for Ashkenazi corrections..

    They are not using a segment based system like 23andMe, and obviously not since FTDNA reports all the segments and requires more than one to make a match to begin with.

    FTDNA has toyed with the idea of using PF data to influence Ashkenazi corrections. Nothing in place as of today.

    .
    Thanks Matt.. I thought you had posted..

    Leave a comment:


  • mkdexter
    replied
    I posted once that FTDNA has some ideas for Ashkenazi corrections..

    They are not using a segment based system like 23andMe, and obviously not since FTDNA reports all the segments and requires more than one to make a match to begin with.

    FTDNA has toyed with the idea of using PF data to influence Ashkenazi corrections. Nothing in place as of today.

    .

    Leave a comment:


  • Kathleen Carrow
    replied
    Originally posted by Javelin View Post
    I doubt they ever will need such a correction because their cutoff for shortest segment is longer than 23andme's, which is an excellent idea. This might eliminate the odd real matches for some people, but it will also prevent many very distant matches, with no hope of finding a paper trail, from clogging anyone's system.
    Oh OK.. I thought I had seen a few people post that they were working on such a correction but had not yet made it operational..

    Leave a comment:


  • Javelin
    replied
    Originally posted by Kathleen Carrow View Post
    Hmmm..that is not actually what I heard them say.. My feeling is that they are trying to resolve the Ashkenazi correction which surely FTDNA will also have to do as their database grows...
    I doubt they ever will need such a correction because their cutoff for shortest segment is longer than 23andme's, which is an excellent idea. This might eliminate the odd real matches for some people, but it will also prevent many very distant matches, with no hope of finding a paper trail, from clogging anyone's system.

    Leave a comment:

Working...
X