An introduction to using the Gephi tool, a popular free network analysis Windows and Mac tool, for analyzing large ICW (In Common With) match networks.
Great tutorial! In a perfect world, the layout would build clusters outward from the center based on shared cM. If it did so, the network diagram would approximate a full circle fan chart.
Not gonna lie. This pushed my brain to a new limit. I love Excel spreadsheets, but this is on an entirely different level. Thank you for this excellent video.
Wow! I am a novice at DNA analysis. I can see this tool has great potential for helping me break out my lines. A bit overwhelming at this point in my learning curve, but I will be back to watch this again when the time is right. Thank you!
Hi Dave, this presentation has changed the direction of my search. I have played with GEPHI couldn't make heaads or tails of the output, but now am on it 'like a rat on a nacho'. I am intrigued by the radial image shown at the 20 minute mark. This 'seems' like it could be a GEPHI product but I wonder f that is so and either way how it was produced. Visually very helpful. Thank you so much for your instruction !!
Thank you for the feedback! The radial wheel is a typical genealogy depiction of someone's ancestry, I just adapted it for the point I was trying to make; that image was generated by hand. You're welcome to the image, I put the blank version at drive.google.com/file/d/15RrFPTDWdjbhMl_oUuW6TXyeCtFSe7vK/view?usp=sharing and the version with color clusters is at drive.google.com/file/d/1Qh7noEsy1SgjrZNxjOzeiC1Fj4Sk6pHt/view?usp=sharing
Gephi 0.9.2 in February 2022 installed with no trouble whatsoever. Download, click, accept defaults and it's there. One negative comment, why don't you format your Excel columns so that values are shown to a consistent number of decimal places? It makes it so much easier to read if decimal points are aligned.
Just starting with Gephi. Perhaps this will become obvious to me later as I start experimenting with Gephi. For now it isn't obvious to me. Can you explain your "2nd File" a bit more. I'm guessing that your SOURCE column is the matchid column from DNAGedcom icw file. I'm guessing that your MATCHID column is the matchname column from DNAGedcom icw file with names privatized? I'm guessing that your TARGET column is the icwid column from DNAGedcom icw file. I'm guessing that your ICWID column is the icwname column in the DNAGedcom icw file with names privatized? Is that correct?
Correct. Gephi requires that the "node file" has a column headed "ID" that gives unique labels to the nodes, so I just used the ids that came out of the DNAGedcom match file. Then Gephi requires that the "edges file" has one column called "Source" and another called "Target" that defines the line between those two nodes. Those columns have to refer to the same labels that the "ID" file did. So for Source and Target in my example I just used the same matchid and icwid that came from the DNAGedcom icw file. In the original DNAGedcom match file I privatized the actual names of the matches, replacing them with "Match #xxxx". Gephi doesn't actually require additional columns besides Source and Target in the "edges file" so I could have just deleted them from that file, but I chose to leave them in just also privatized. But yes those are the matchname and icwname columns originally; I'm honestly not sure how the column headers got switched to "matchid" and "icwid" from their originals in the DNAGedcom file, but Gephi doesn't really care what any additional columns are called.
What is your NUMBERING SYSTEM called? I couldn't find anything like it on the internet. The closest I found was the d'Aboville System... I tried imitating it in a spreadsheet but your mistake on Magaret M (1883-1964) threw me and I got hopelessly confused the further down I went. Would Binary Ahnentafel numbering not work for your Gephi labels?
Hi - thanks for the catch on Margaret M - she should be 1-2-2. The numbering scheme is described in the video starting about timestamp 9:33; it is certainly based on other systems like Ahnentafel but this worked for me because it was easy to see the number of generations back and the genders of the generations all at once in the labels on the nodes so nodes from different ancestors along the same ancestral paths were more obvious. But it’s not crucial that you adopt the exact same method - any system that works for you would be fine!
Ciao sono Kevin io ho fatto test myheritage e ho parecchi gruppi marchese ma uno in particolare e sono tutti dai Balcani qualcuno tra loro e più vicino e anche tra loro ci sono delle distanze e poi in mezzo ha loro ci sono io e un po' strano.io sono sud Italy puglia Campania basilicata.e loro hanno di sicuro origini rom .io nel mio lato materno mi manca la linea di mia nonna che non si sa chi era il padre e la linea della bisnonna che madre della mia nonna che ho detto .anche lei era senza padre era figlia di padre incerto.tu credi che ci sia una connessione?perché nel mio albero e lì che manca qualcosa.poi ho controllato i segmenti stessa posizione dei segmenti che son che sono paterni e non triangolano.allora può darsi che siano materno.anche se hanno tra due segmenti e tre segmenti e un segmento.ma solo uno segmento di loro triangola con me.
![Lp. Последняя Реальность #107 РОДНОЙ ДОМ [Финал] • Майнкрафт](http://i.ytimg.com/vi/IK3QKzKUlHM/mqdefault.jpg)
Great tutorial! In a perfect world, the layout would build clusters outward from the center based on shared cM. If it did so, the network diagram would approximate a full circle fan chart.
Not gonna lie. This pushed my brain to a new limit. I love Excel spreadsheets, but this is on an entirely different level. Thank you for this excellent video.
Fabulous video and explanation of how to use this program for genealogy. Thanks so much!
Finally, a visual person who uses dna on Gephi - thank you!
Thank you Dave! This video is just what I needed to analyze the gephi match network I've generated.
Excellent tutorial I used to analice my Ancestry matches! Thanks
Wow! I am a novice at DNA analysis. I can see this tool has great potential for helping me break out my lines. A bit overwhelming at this point in my learning curve, but I will be back to watch this again when the time is right. Thank you!
Thanks for sharing!! This is great.
Thanks Dave for an excellent video. Great detail and easily replicable.
Hi Dave, this presentation has changed the direction of my search. I have played with GEPHI couldn't make heaads or tails of the output, but now am on it 'like a rat on a nacho'. I am intrigued by the radial image shown at the 20 minute mark. This 'seems' like it could be a GEPHI product but I wonder f that is so and either way how it was produced. Visually very helpful. Thank you so much for your instruction !!
Thank you for the feedback! The radial wheel is a typical genealogy depiction of someone's ancestry, I just adapted it for the point I was trying to make; that image was generated by hand. You're welcome to the image, I put the blank version at drive.google.com/file/d/15RrFPTDWdjbhMl_oUuW6TXyeCtFSe7vK/view?usp=sharing and the version with color clusters is at drive.google.com/file/d/1Qh7noEsy1SgjrZNxjOzeiC1Fj4Sk6pHt/view?usp=sharing
Thank you for this video! I found it extremely helpful.
Gephi 0.9.2 in February 2022 installed with no trouble whatsoever. Download, click, accept defaults and it's there. One negative comment, why don't you format your Excel columns so that values are shown to a consistent number of decimal places? It makes it so much easier to read if decimal points are aligned.
Just starting with Gephi. Perhaps this will become obvious to me later as I start experimenting with Gephi. For now it isn't obvious to me. Can you explain your "2nd File" a bit more.
I'm guessing that your SOURCE column is the matchid column from DNAGedcom icw file.
I'm guessing that your MATCHID column is the matchname column from DNAGedcom icw file with names privatized?
I'm guessing that your TARGET column is the icwid column from DNAGedcom icw file.
I'm guessing that your ICWID column is the icwname column in the DNAGedcom icw file with names privatized?
Is that correct?
Correct. Gephi requires that the "node file" has a column headed "ID" that gives unique labels to the nodes, so I just used the ids that came out of the DNAGedcom match file. Then Gephi requires that the "edges file" has one column called "Source" and another called "Target" that defines the line between those two nodes. Those columns have to refer to the same labels that the "ID" file did. So for Source and Target in my example I just used the same matchid and icwid that came from the DNAGedcom icw file.
In the original DNAGedcom match file I privatized the actual names of the matches, replacing them with "Match #xxxx". Gephi doesn't actually require additional columns besides Source and Target in the "edges file" so I could have just deleted them from that file, but I chose to leave them in just also privatized. But yes those are the matchname and icwname columns originally; I'm honestly not sure how the column headers got switched to "matchid" and "icwid" from their originals in the DNAGedcom file, but Gephi doesn't really care what any additional columns are called.
What is your NUMBERING SYSTEM called? I couldn't find anything like it on the internet. The closest I found was the d'Aboville System... I tried imitating it in a spreadsheet but your mistake on Magaret M (1883-1964) threw me and I got hopelessly confused the further down I went. Would Binary Ahnentafel numbering not work for your Gephi labels?
Hi - thanks for the catch on Margaret M - she should be 1-2-2. The numbering scheme is described in the video starting about timestamp 9:33; it is certainly based on other systems like Ahnentafel but this worked for me because it was easy to see the number of generations back and the genders of the generations all at once in the labels on the nodes so nodes from different ancestors along the same ancestral paths were more obvious. But it’s not crucial that you adopt the exact same method - any system that works for you would be fine!
Ciao sono Kevin io ho fatto test myheritage e ho parecchi gruppi marchese ma uno in particolare e sono tutti dai Balcani qualcuno tra loro e più vicino e anche tra loro ci sono delle distanze e poi in mezzo ha loro ci sono io e un po' strano.io sono sud Italy puglia Campania basilicata.e loro hanno di sicuro origini rom .io nel mio lato materno mi manca la linea di mia nonna che non si sa chi era il padre e la linea della bisnonna che madre della mia nonna che ho detto .anche lei era senza padre era figlia di padre incerto.tu credi che ci sia una connessione?perché nel mio albero e lì che manca qualcosa.poi ho controllato i segmenti stessa posizione dei segmenti che son che sono paterni e non triangolano.allora può darsi che siano materno.anche se hanno tra due segmenti e tre segmenti e un segmento.ma solo uno segmento di loro triangola con me.