One possible explanation: If you are using v3 data, there are ~ 1,000,000 SNPs to use when you're comparing M kits with M kits. A kits have only 700,000 SNPs available for comparison, so it's harder to reach the SNP threshold. But I'm puzzled that there was a reduction from your original M kit to the fixed M kit. That implies that at least some of the fixes created opposite homozygotes between potential matches. Could you do a few one-to-one comparisons of the dropouts with different SNP and mismatch thresholds? Ann Turner On Wed, Dec 16, 2015 at 12:41 PM, David Schroeder via < genealogy-dna@rootsweb.com> wrote: > Continuing with my analysis of raw data files that have been 'fixed' for > no-calls using SNP information from matching RSIDs found in 23andme and > AncestryDNA. > > I uploaded the fixed files to gedmatch. My original kit from my 23andme was > already in place. I used the Tier 1 utility, "Matching Segment Search." I > used a minimum segment size of 5 cm without graphic. I downloaded the > matching list of segments into an Excel Spreadsheet. I used the segment > matches as I feel I get better information than the "one-to-many." > > Here are my results: > > 1925 gedmatch_segment_matches_M080859 (original 23andme kit with a 2.2 % > error rate) > > 1741 gedmatch_segment_matches_M306764 (fixed 23andme kit with a 1.5 % > error > rate- about a 10% reduction in segment matches) > > 1263 gedmatch_segment_matches_A146269 (fixed Ancestry kit with a 1.0% > error > rate - I haven't uploaded my original AncestryDNA file) > > I do have issues with a lot of pileups on chromosome 15: > > Number of Segment Matches on Chromosome 15 - large number of pileups > Original 23andme: 603 > Fixed 23andme: 482 > Fixed Ancestry: 197 > > Compared to number of Segment Matches on Chromosome 6: > Original 23andme: 75 > Fixed 23andme: 82 > Fixed Ancestry: 84 > > It is an interesting situation. I expected fewer matches on my fixed kit, > but I did not expect so much less on my ancestry kit. Any ideas why? I am > going to dig deeper into the data. > > David > > -----Original Message----- > From: David Schroeder [mailto:dschroed991@sbcglobal.net] > Sent: Monday, December 14, 2015 10:48 PM > To: 'genealogy-dna@rootsweb.com' > Subject: RE: [DNA] My Raw Data Files - Comparison 23andme vs AncestryDNA > > I was able to 'fix' the no-calls for matching RSIDs on both Ancestry and > 23andme when one, or the other, was not a no-call. I fixed 6,632 on 23andme > and 6,708 on Ancestry. > > Interestingly enough, there were 3,833 that were left as no-calls on both > 23andme and AncestryDNA for the same RSIDs. I am wondering if these are the > result of particularly difficult locations to test, or perhaps the SNP is > rare in my genome? The tests were over two years apart. > > I uploaded both fixed raw data files to gedmatch to see how it may affect > my > 'one-to-many' matches. (Will have to wait on the processing). I ran the > Gedmatch File Diagnostic Utility, and the fixed files had significantly > reduced my error rates. It seems that most of my errors are in the X, Y or > MT Chromosomes. > > David > > > > ------------------------------- > To unsubscribe from the list, please send an email to > GENEALOGY-DNA-request@rootsweb.com with the word 'unsubscribe' without > the quotes in the subject and the body of the message >