igraph-help
[Top][All Lists]

## Re: [igraph] help on assortativity HLA

 From: Fernando Barraza Subject: Re: [igraph] help on assortativity HLA Date: Thu, 8 Aug 2019 10:22:22 -0500

Hi Antoine,

If I understand correctly, your last part of code is estimating the assortativity as if individuals that matches tend to links to the other that also matches and the same with the unmatches. This is correct?

I am some suggestions on how to calculate assortativity when we have a scalar characteristic with multiple values on each vertex, but I want to be sure if these suggestions could apply for your problem.

Best regards,

Fdo.

El dom., 28 jul. 2019 a las 13:48, Chaillon, Antoine (<address@hidden>) escribió:
Dear Colleagues,
I am using igraph  to explore assortativity of HLA profiles between linked individuals (HIV)
My ‘problem’ is that for each HLA group (A, B or C), I have 2 values associated with each vertex .
My goal is to determine if :
HLA A1 OR A2  are more likely to be ‘assortative’ in the network
HLA B1 OR B2  are more likely to be ‘assortative’...
HLA C1 OR C2  are more likely to be ‘assortative’...

For now, I compare it to a random distribution.
For example with HLA C (C1 or C2), I have 28 unique HLAC profile.
In a network with 286 vertices, First I calculate the rdm_assortativity as follow:
RandomSample <- sample(c(1,2,5,4,6,7,8,9,10,3,11,12,14,13,16,17,15,18,19,20,21,22,23,24,25,26,27,28), 286, replace=TRUE) ## my guess is that the dimension is incorrect because it does not take into account the 2 possibilities (C1 and C2)
rdm_assortativity=assortativity_nominal(graph=ALLEDGES2, types=RandomSample, directed = F)

Next, I want to determine the observed assortativity. What would you suggest?
For now, I manually created an additional column ‘HLAC_match’ with 1 for ‘match’ and 2 for ’no match’ and then calculate the assortativity as follow:
assortativity_nominal(graph=ALLEDGES2, types=HLAATTRIBUTES\$HLAC_match, directed = F) # where ALLEDGES is my edge list

Does that make sense?
Any thoughts?
thank you in advance!
a

_______________________________________________
igraph-help mailing list