Image 1. Crime boss reveled through network analysis. This is a criminal network from the show, “Breaking Bad” showing the known entities (photographs) and their associations (lines). If you watched the show, you know that Gustavo Fring (middle photo) was the mastermind behind a large methamphetamine enterprise. This network illustrates this role -Gustavo is the highest connected individual in the network.
A criminal enterprise is basically a community of interacting individuals with specific functions. Visualizing the individuals and their associates as a network is one of the best ways to understand an organization’s structure and thereby dismantle it. For example, you can tell from the Breaking Bad criminal network above (Image 1) that Gustavo Fring is probably very influential. Gustavo’s removal is likely to shatter the organization into tiny headless pieces because he is highly connected, and the junction that connects several disconnected elements such as Gale and Hector. Each of us, criminal or not, have communities that function in similar ways.
Image 2: Hypothetical social/professional network. The woman in the center has the most connections and may be very important. Note that everyone in the network is no more than just 1 person away from meeting her. Its a small world isn’t it?
The structures of our social and professional networks also have important information weaved into them. A study in the Journal of Corporate Finance demonstrated that CEOs with a diverse social network can have a direct impact on the success of an organization. Another provocative article extrapolated from a study (Statistics in Medicine) that “people from the majority group and men” have immense power in their social and professional networks that can be used to do work for social justice in two important ways.
Speaking up about discrimination and implicit bias to your associates (parents, friends, coworkers, training partners, etc) will reach the associates of your associates
Diversifying your network can be very empowering for everyone on both sides. More importantly, establishing a relationship with a person from a different culture forms a fragile junction that joins two otherwise isolated communities.
Diversity adds value to a community, and the value of a network scales with the number of individuals. What could you learn about yourself and others by inventorying your social network and the social networks of the associates in your network? How many people are you away from meeting a local celebrity or president Barak Obama?
Networks are not limited to Homo sapiens. A network can help us better understand the structure of any community of living organisms. Microscopic organisms are no different because we know they don’t live in isolation. Instead, microbes function within densely populated, highly interactive communities where each species plays a specific role. Constructing the social networks of microbial communities across space and time has the potential to improve our understanding of microbial community structure, the nature of individual interactions, and eventually uncover the species who’s removal would result in fragmentation of the network -keystone species (sensu Paine 1969).
I learned from some work I conducted at station ALOHA that the single-celled eukaryotes (protists) living in the water column from 5m to 770m formed three communities according to depth, and that seasonal changes in community structure was not significant. In other words, the majority of the protistan community in the North Pacific Subtropical Gyre is the same in the spring, summer, and winter, but changes down the water column at different specific depths. The changes appeared to be most influenced by the sharp decrease in sunlight intensity.
Image 3: Vertical distribution of protistan groups in the North Pacific Ocean (NPSG). There are a diversity of preferences, mostly based on sunlight, which does not penetrate beyond 175m (dotted line).
I created a sparse ecological network* of positive interactions from this dataset to investigate the community-level connectivity throughout the water column. The samples did not form one single network. Instead, the samples bifurcated into sun-lit (photic) and aphotic subnetworks (Image 4).
We guessed that Gustavo Fring was was the crime boss because he had more connections than the average individual in the association network (Image 1 above). Such an approach may or may not neatly apply to microbial communities. The distribution of the number of connections in the microbial network follows what we saw in the Breaking Bad network: lots of individuals with very few associates, less individuals with more than average number of associates, and a tiny fraction of individuals with a disproportionately high number of associates. If all networks are equal, and that’s a gigantic if, these individuals are like the ‘crime bosses’ of this microbial community because their removal could fragment the community. The hard part is explaining how, and proving it?
Image 4: Protistan interaction network from 5m to 770m. Circles represent an individual species, the size of the circle represents the number of interactions (lines) it has with other species. Circle colors represent the depths the species was found in abundance.
This story seems neatly packaged to the untrained eye, but constructing the network was the easy part. Despite the sophisticated statistics, algorithms, and carefully collected and pruned data used to construct the network, these results are only a guess; my Hypothesis. The hard part will be finding the proof.
I’ve stated to address the following questions:
Who are the organisms and their associates in the network?
What is already known about the interactions between these organisms from culture-based observational studies?
Do other methods generate the same results?
These and many other questions are just the tip of the iceberg, but are necessary to produce real knowledge about the nature of microbial community structure and the utility of network analysis to reveal properties that hold up in the real world. I see promise in this method as I move forward.
As always…
