Thoughts from a wildlife biologist

[TomAug85]

Hey folks,

I thought I should represent whole organism biologists here as it seems the world of networks and to some extent Gephi is dominated by physicists, computers scientists and anthropologists.

Networks analysis, and specifically social networks analysis is an up and coming field in wildlife biology. These networks help us understand how wild populations interact naturally, how these interactions are effected by human actions, how diseases spread in these populations and how social organisation is reflected spatially.

Gephi could be a powerful tool in this field, but there are a few places it needs improvement.

The first is the inability to represent node attributes by different shapes. Animal networks are made up of males and females and different age classes which are usually of functional interest and so need to be visualised separately. This fundamental function, not yet in Gephi (and I would suggest should be a high, not medium, priority), is the main reason why I, and I imagine others, do not yet use Gephi in our peer reviewed publications. I won't pretend to understand why this might be a complicated implementation.

Gephi can now accept .VNA files, the output of UCInet, Netdraw and SocProg, software commonly used by wildlife biologists. However this does not seem to work with my data, I'm happy to supply this to developers to help fix this bug.

Biologists often use bipartite networks. These represent the interactions between two node classes, usually predator prey, pollinator-plant, or individual and roost/nest site. I have been able to make these by hand but suspect a layout could be designed to do this more easily by fixing each class of node to a specific y-coordinate.

Other potential plugins include modelling disease spread through a network and the comparison of empirical networks to random permutations of the same data to test significance.
Whilst I know about animal ecology and social networks I have a limited background in programming. I'd be happy to talk to people more about how Gephi could be developed to support this growing demand. Perhaps in doing so we will see a growing number of Gephi networks in peer reviewed journals rather than blogs.

Regards,

Tom (PhD Student in Social Networks of Wildlife)

[admin]

Hey, you forgot Formula One fans

Anyway I'm glad to read your thoughts. We try to provide a generic tool but our minds are of course shaped by our own usages in cartography creation.

The first is the inability to represent node attributes by different shapes. Animal networks are made up of males and females and different age classes which are usually of functional interest and so need to be visualised separately. This fundamental function, not yet in Gephi (and I would suggest should be a high, not medium, priority), is the main reason why I, and I imagine others, do not yet use Gephi in our peer reviewed publications. I won't pretend to understand why this might be a complicated implementation.

A new visualization engine is on development, and it will be possible.

Gephi can now accept .VNA files, the output of UCInet, Netdraw and SocProg, software commonly used by wildlife biologists. However this does not seem to work with my data, I'm happy to supply this to developers to help fix this bug.

Please report a bug with your file attached, and we'll find what's wrong.

Biologists often use bipartite networks. These represent the interactions between two node classes, usually predator prey, pollinator-plant, or individual and roost/nest site. I have been able to make these by hand but suspect a layout could be designed to do this more easily by fixing each class of node to a specific y-coordinate.

I'm interested to know more on what is useful to study bipartite networks. Could you start another thread dedicated on it so we can discuss the specifications, missing layouts and algorithms? We need to express the needs clearly so that developers can implement them.

Other potential plugins include modelling disease spread through a network and the comparison of empirical networks to random permutations of the same data to test significance.

Sounds like a plug-in the core developers won't implement. You should find someone ready to help you to code it.

Whilst I know about animal ecology and social networks I have a limited background in programming. I'd be happy to talk to people more about how Gephi could be developed to support this growing demand. Perhaps in doing so we will see a growing number of Gephi networks in peer reviewed journals rather than blogs.

Sometimes quick hacking and dirty engineering is better at capturing the reality than the science process do. Thank you for allowing me to end my post with a total troll, as I'm a PhD too

[TomAug85]

Thank you for all your feedback, I'll take actions as you recommended.

Not sure what a troll is but I would say that if your technique better explains the data it doesn't matter if it is rough and ready or pain-staking constructed, it is still the better scientific method.

Tom

[TomAug85]

Re: the .VNA issue I have discovered this is because Netdraw outputs comma delimited files. Could there be an option in Gephi to support this?

[admin]

Hi,

Don't hesitate to report the bug on Launchpad so that we can process it

thanks!