Glen, I like your idea. Maybe an additional feature of the collision code you describe would be to provide a confidence factor based on the quantity of DNA allele matches and the supporting tree structure. A confidence factor could be included for all projected values along a path to a Most Recent Common Relative. Additionally, if confidence factors are included, the system could project ancestor DNA beyond a common intersect. Obviously, the confidence would drop with each previous generation but it would add an interesting facet to the analysis.
Confidence factors would be an easy way to verify that the recorded genealogy records conform with the projections of the DNA analysis. I currently have ancestors in my genealogy records that don't correlate to my DNA testing. It would be nice to have a tool that could be use to show fellow researchers where these disconnects occur. The two most common disconnects are either; a documented common relative when DNA analysis doesn't show one exists or DNA analysis identifies a common relative that was previously unknown. The former could be red-flagged and the genealogy records marked as suspect that are associated with the unproven connection. The later could be an information pop-up notifying the user that a common ancestor probably exists (in a probable time range) that is not in their genealogy records. Carrying your tree walker concept one step farther, would be to allow the user to pick an ancestor (maybe with projected DNA results) and let the system walk the tree to give you the names of potential candidates for additional DNA testing. It could even sort them in the order that would give you the best DNA coverage for a particular ancestor. This could be very beneficial for anyone wanting to identify a list of contacts for additional testing or wanting to know what branches of their family tree need to be expanded to find potential candidates. Everyone know that DNA analysis doesn't work without genealogy records as a base from which to work. Combining the two disciplines into one tool is an obvious benefit. Like most new things it is difficult to see all the benefits and possible uses until there is an opportunity to explore these possibilities more. Legacy's attempt to add DNA recording is commendable. It lets us have the opportunity to start recording DNA values (the obvious first step) while letting us workout how this data can best be analyzed to provide the most beneficial information. Steve On 10/23/05, Glen Todd <[EMAIL PROTECTED]> wrote: > > I've already registered my complaints on the V6 'DNA' feature - while the > implementation is competent, that's of little value when the design is so > badly though out. As a result, it's basically useless. > > Having said that, I've been thinking about it (I'm a software engineer > myself and can't resist playing with such things). One of the > characteristics of DNA information that make it different from most > genealogical record information, is that it applies not only to the > individual tested but to that individual's entire Y-line or mt-line (with > exceptions that I'll get to later). Initially this seems as if it would > be a whole new ball game, but on inspection it's hardly unique in the Legacy > world. Similar relationships exist with 'master records' such as surnames, > geo locations, and references, with many to one relationships between > individual records and master records. That's the way that I'd handle > this; when mt- or Y- information is entered, a master record would be > created and a link to that master record inserted into the individual > record. The link would also contain related information such as whether > the link was tested or projected. Then because of the way that mt- and > Y-DNA work, a tree walker in the code could follow that line back and insert > 'projected' links in the other records for that line. (Mutations, which > are rare) would be handled by a special connector between master records.) > When the tree walker detects a collision with another projection (trivial to > implement) a chunk of collision resolution code comes into play. If both > projections reference the same master record (most probable result), there's > no problem. Different records with identical signatures should in some > cases be merged, but I would definitely pop up a message box first. If > the patterns DON'T match, that means that there's one of two types of > anomalies; a small (1 or 2) value difference at one marker is likely a > mutation, while a significant difference at several markers would most > likely be a paper trail error such as a mis-attribution or a false paternity > event, and should be flagged as an error needing resolution. This > complicates the tree walker code a bit, since in the event of something like > an adoption it would have to look at both sets of parents, but that's not > very much of a complication as long as the designer is aware of the > requirement. This could also be integrated into the 'Research Guidance', > since matching patterns appearing in different lines (especially close to > the same place and time) might invite further consideration. A tree > walker following a line back to a solid triangulation could also set or > validate the 'Biological' child status flag. > > Well, you get the idea. > > Glen > > Legacy User Group Etiquette guidelines can be found at: > http://www.LegacyFamilyTree.com/Etiquette.asp > > To find past messages, please go to our searchable archives at: > http://www.mail-archive.com/legacyusergroup%40mail.millenniacorp.com/ > > To unsubscribe please visit: > http://www.legacyfamilytree.com/LegacyLists.asp > -- Tell your friends and family about our Winslow Family Y-DNA project! They may know a Winslow who would like to participate in this exciting project. Legacy User Group Etiquette guidelines can be found at: http://www.LegacyFamilyTree.com/Etiquette.asp To find past messages, please go to our searchable archives at: http://www.mail-archive.com/legacyusergroup%40mail.millenniacorp.com/ To unsubscribe please visit: http://www.legacyfamilytree.com/LegacyLists.asp
