Certainly a proportional-area Venn diagram is simple, but it will only show the amount of overlap among the inventories--not the individual points unless you actually draw them uniformly within their allowed space. Your request about being able to use properties such as e.g. molecular weight for a 2D plot is mutually exclusive with a Venn layout, as the position of each pot would be dependent on the properties and not necessarily coincide with the required position within the Venn overlaps.

If you want to describe the positions of the compounds in 2D or 3D space based on their properties, possibly the only way you'd have to "group" them would be to differentiate the dots themselves. You could certainly use combined colors as you suggest (e.g. call your lists "R", "G", and "B", and color each dot accordingly: E, G, B, RG, RB, GB, RGB). Alternatively, you could do what we used so often for printing in the striped-paper era: Set a symbol for each list, and combine symbols for the combinations (e.g. A is square, B is circle, then AB is square within circle).

Should parameter-mandated coordinates be secondary, then you could go back to Venns. Exact solutions do not exist for three-group Venns using circles or squares, but you could do it with ellipses or--better yet--rectangles using e.g. DrawVenn. The advantage of overlapping rectangles is that you could define regular tiles within the Venn, and assign one tile to each compound. The amount of tiles would be the number of distinct compounds you'd have in the pooled lists, and the area of each tile should be the total (combined) Venn area divided by the number of tiles (=number of distinct compounds in the pooled list). The position of each tile (=each compound) within its corresponding overlap (or exclusive) area would be arbitrary.

Now, if you still need to represent a parameter (e.g. molecular weight), instead of choosing homogeneous, naturally-mixed colors for the Venn areas (see e.g. figure 1 in the attached paper), you could shade each tile according to that parameter. For example, let's suppose you want to represent molecular weight. Once you have your tiled Venn, if e.g. the tile (=compound) belongs to the BG area (that is, yellow), you'd use a yellow ramp from light to dark yellow to represent mol weight, whereas tiles in the exclusive portion of the B area (that is, compounds only present in the B list) would be light to dark blue. You could use grayscale for the RGB area.

A second parameter (or even the first one) could also be added as height. I can imagine a landscape where each compound could be a floating semitransparent cube within the "tower" defined at the base by the overlapping areas of the Venn diagram.

If you prefer the appeal of overlapping ellipses (or even circles, although this will be only approximate) you could draw their limits and use bid (colored) dots uniformly distributed within the ellipses as well.

Good luck with your visualization! 

Article The biodiversity data knowledge gap: Assessing information l...

A mathematically interesting q. Just a hunch..Data maybe converted to pixel values and segmentation / levelset  plugins of imagej might help in visualization.

Good Luck

Cheers

Another possibility is to use machine learning .see

1.http://cs.stanford.edu/people/karpathy/convnetjs/demo/classify2d.html

2.https://www.tensorflow.org/

Cheers

Thanks Narasim. I like the idea of converting data to pixels but that's too general. I will most probably use the hiveplot.