Comments on: Thesis: 1.4 Network datasets http://toreopsahl.com bouncing ideas Tue, 24 Aug 2010 19:03:46 +0000 hourly 1 http://wordpress.com/ By: Tore Opsahl http://toreopsahl.com/publications/thesis/thesis-14-network-datasets/#comment-770 Tore Opsahl Thu, 07 Jan 2010 16:43:01 +0000 http://thetore.wordpress.com/?page_id=285#comment-770 Dear Tiha, Thank you for taking an interest in my blog, and thesis. I think this is a reflection of a more general issue that network scientists have to deal with sooner or later. Complete networks do not exist in isolation, but form part of a wider system. In fact, it is similar to ego networks that are part of a larger complete network. It is easy to say that we simply define our network as the 306 neurons without taking into consideration the system around them. However, we are then missing a key feature: certain neurons are connected to the body while others are not. By not having information about connections outside the network, we are not able to characterise the nodes fully by simply looking at the network. By looking at the last 7 nodes in the network, it is easy to see that there are "pendant" nodes that simply has one weak connection to someone else (sensory neurons possibly?), and some nodes that are receivers of many ties without having any outgoing ties. As you mention, 134 neurons are connected to node #305 (incoming ties) through 1700 synapses or gap junctions. This is the maximum in-degree and in-strength of any node in the network. As a layman within the neural network research, I would assume neuron #305 is a major connector to the rest of the worm. Best regards, Tore [sourcecode type="text"] node degree.out output.out degree.in output.in 300 1 1 0 0 301 1 1 0 0 302 1 1 0 0 303 0 0 10 13 304 0 0 0 0 305 0 0 134 1700 306 0 0 31 68 [/sourcecode] To get this table: [sourcecode type="text"] # Load tnet library(tnet) # Load network celegans.n306.net <- read.table("http://opsahl.co.uk/tnet/datasets/celegans_n306.txt") # Run out- and in-degree function out <- data.frame(degree_w(celegans.n306.net), degree_w(celegans.n306.net, type="in")[,2:3]) # Name columns dimnames(out)[[2]] <- c("node","degree.out","output.out","degree.in","output.in") # Output results out[300:306,] [/sourcecode] Dear Tiha,

Thank you for taking an interest in my blog, and thesis.

I think this is a reflection of a more general issue that network scientists have to deal with sooner or later. Complete networks do not exist in isolation, but form part of a wider system. In fact, it is similar to ego networks that are part of a larger complete network. It is easy to say that we simply define our network as the 306 neurons without taking into consideration the system around them. However, we are then missing a key feature: certain neurons are connected to the body while others are not. By not having information about connections outside the network, we are not able to characterise the nodes fully by simply looking at the network.

By looking at the last 7 nodes in the network, it is easy to see that there are “pendant” nodes that simply has one weak connection to someone else (sensory neurons possibly?), and some nodes that are receivers of many ties without having any outgoing ties. As you mention, 134 neurons are connected to node #305 (incoming ties) through 1700 synapses or gap junctions. This is the maximum in-degree and in-strength of any node in the network. As a layman within the neural network research, I would assume neuron #305 is a major connector to the rest of the worm.

Best regards,

Tore

node degree.out output.out degree.in output.in
 300          1          1         0         0
 301          1          1         0         0
 302          1          1         0         0
 303          0          0        10        13
 304          0          0         0         0
 305          0          0       134      1700
 306          0          0        31        68

To get this table:

# Load tnet
library(tnet)

# Load network
celegans.n306.net <- read.table("http://opsahl.co.uk/tnet/datasets/celegans_n306.txt")

# Run out- and in-degree function
out <- data.frame(degree_w(celegans.n306.net), degree_w(celegans.n306.net, type="in")[,2:3])

# Name columns
dimnames(out)[[2]] <- c("node","degree.out","output.out","degree.in","output.in")

# Output results
out[300:306,]
]]> By: Tiha http://toreopsahl.com/publications/thesis/thesis-14-network-datasets/#comment-767 Tiha Thu, 07 Jan 2010 04:49:13 +0000 http://thetore.wordpress.com/?page_id=285#comment-767 With reference to the C. elegans graph: the in-degree of node 305 is about 130; its out-degree is zero. What's a neuron doing all day with no out-degree and such a vast source of input signals? I have tried to find some information on this but have failed to dig up anything so far. I assume that 0 in-degree could mean a sensory neuron and 0 out-degree some sort of an "actuator;" that is their inputs and outputs respectively are relayed from/to non-neuron cells but that still doesn't solve the #305 mystery. Any thoughts? Kind regards, With reference to the C. elegans graph: the in-degree of node 305 is about 130; its out-degree is zero. What’s a neuron doing all day with no out-degree and such a vast source of input signals? I have tried to find some information on this but have failed to dig up anything so far.

I assume that 0 in-degree could mean a sensory neuron and 0 out-degree some sort of an “actuator;” that is their inputs and outputs respectively are relayed from/to non-neuron cells but that still doesn’t solve the #305 mystery. Any thoughts?

Kind regards,

]]>