Greetings!
I have an issue that drives me crazy this evening...
I have a list of gene vectors, downregulated in different transgenic plants and I want to make a Venn diagram to visualize it and to show the intersections between plants.
But! The results from any package I used (in R) gaves me something like this (the uploaded picture 1)...
What's bothering me:
1. The numbers on "clear" (not intersected) parts of a diagram are lower, than the gensets I have. And I tried to use factor instead of character vectors, to remove possible duplications, to remove symbols (like space) that could cause software misunderstanding - all gaves me nothing... same result.
2. The intersection of vectors is not true - on the picture you can see that the intersection of 2 datasets (of 365 and 154 genes) - is 1133 genes!! How could that be?
The manual usage of intersect function on the same dataset gaves pretty correct results.
Maybe I am misunderstanding about Venn diagrams? Because in a web I found many examples of such strange mistakes - on the second picture from Datanovia you can see that the intersection of the red elliplse (of 58) and yellow (of 144) is 66!
It seemes logical to me that the intersection of 2 vectors cannot be greater than the length of a smaller vector. What am I doing wrong or misunderstanding?
Dear Igor V Gorbenko ,
I think I understand the logic of the second example you gave. It depicts four sets of data (A, B, C and D). See https://www.datanovia.com/en/blog/beautiful-ggplot-venn-diagram-with-r/ If you look at the (simple) enclosed picture for example (dataset A and B) you see that the A contains 300 (149+151) and B contains 525 (374+151) datapoint. The number of ‘unique’ genes (here) is 674. The overlap between A and B is 151. 149 and 374 means not the size of the sector but the number of datapoints without overlap.
In other words, the numbers 58 and 144 are not telling you the size of the set but ‘simply’ the number of datapoints with no overlap to other datasets.
Hope this clarifies your mystery somewhat.
Best regards.
I believe Rob is correct.
Since you are using the intersect function, the numbers in your figure (e.g. 365 and 154) are the number of genes without any intersection.
The total genes of each set (e.g. OE21) will be the sum of all the numbers in each intersection + genes with no intersection. I couldn't do full sum for you as the core intersection number is missing.
Dear Ryan Gourlie and Rob Keller , thanks a lot! It resolves one question for me, but another one stays unclear...
Here I am talking about the first uploaded picture. If I am using the intersect() function manually - I get that the overlap between datasets OE7 and OE21 is about 1400 genes (and the intersection on the diagram shows 1133 genes). The overlap between OE_0 and OEM15 (which is the most critical in the study) is 41 genes - and the diagram shows 27 genes.
Does it mean that overlap of the sections shows genes (27) that are exclusievely in this particular overlap? And to get the full number of overlapped genes between 2 samples I need to sumhow summarize other overlaps?
Dear Ryan Gourlie and Rob Keller , yep, I finally get it. To get the full number of intersected genes I need to summarize numbers from all the zones that are in intersect of 2 figures - not just the zones of their "pure" intersection without overlap with others. Thank you very much!
- Badges
- Science topic
- Similar topics
- Biological Science
- Immunology
- Antibodies