Andreas, just a question because I'm uncertain about some things. I was unaware that any of the testing companies that allow the download of raw DNA data supply anything but the actual SNP values as digitized by the Illumina iScan system...whichever BeadChip version is in use and whatever custom programming is allowed (e.g., the GSA chip allows for up to 50,000 lab-configured bead types). The sequencing is established by the Illumina hardware, microarrays, and HTS workflow. At the end of that workflow, after enzymatic extension, fluorescent staining, and imaging on the iScan, Illumina software automatically performs some analysis and genotyping for quality control. But the raw data comes out the same regardless of the direct-to-consumer company selling and running the test, doesn't it?
In other words, as examples, the "stitching" imputation that MyHeritage does and Ancestry's use of "Underdog" genotyped phasing and "Timber" match-filtering algorithms are all applied computationally to the raw data, but don't alter the raw data itself? I know that last genotyping step that Illumina performs involves imputation for low-frequency and common minor allele variants, but I believe that's beyond the control of the testing companies.
Doesn't that mean--since the raw data is being uploaded to and used in GEDmatch, which does no stitching or imputation (I can't count things like SNP mismatch bunching limits or SNP density or the correlations necessary to align the differing versions of the human genome reference maps)--that what we see on GEDmatch is completely unaffected by whatever algorithms the testing companies choose to apply to the data after sequencing for their own reporting purposes?
And I agree with you about the value of triangulated segments, certainly beyond around 2C1R or 3C. But I think Kitty agrees with you also. She even said that functionality was almost ready. Meanwhile, I don't see that Kitty's new utility has any less value than any other "in common with" tool out there, and I believe it adds additional functionality to GEDmatch. I applaud her for it.
