From: Yuchen Yuan (ycyuan_at_bu.edu)
Date: Tue Aug 20 2019 - 20:35:51 CDT

Dear VMD users,

Recently I am using the NetworkView plugin in VMD to do the network and
community analysis. Both in the tutorial and my own systems, the suboptimal
path lengths are a couple of hundred. According to the definition of edge
length, w = -log|c_ij|, where c_ij is the normalized covariance between
node i and j, the path length is simply the sum of w along a particular
path. Since c_ij ranges from -1 to +1, it would not be reasonable to have a
path length larger than 100. And the path length is also impossible to
always be an integer. So I am really confused now.

The following is part of an output of subopt in my systems and the path
lengths are in the parenthesis:

The shortest distance is here 385
The edge length offset is 20
Suboptimal paths will be no longer than 405
The sources are: 62
The targets are: 248
The final paths are:
62, 83, 85, 143, 148, 152, 249, 251, 248, (385)
62, 84, 88, 143, 148, 152, 249, 251, 248, (401)
62, 84, 139, 143, 148, 152, 249, 251, 248, (402)
62, 84, 140, 143, 148, 152, 249, 251, 248, (400)

I didn't find a paper describing the implementation details of NetworkVeiw
so I assume it follows the same protocol as described in the following
papers:

1. Dynamical networks in tRNA:protein complexes. Anurag Sethi, John Eargle,
Alexis A. Black, Zaida Luthey-Schulten, Proceedings of the National Academy
of Sciences Apr 2009, 106 (16) 6620-6625
2. Exploring Residue Component Contributions to Dynamical Network
Models of Allostery. Adam T. VanWart, John Eargle, Zaida Luthey-Schulten,
and Rommie E. Amaro, J. Chem. Theory Comput. 2012, 8, 2949−2961

For example, in the table 1 of ref. 1, some shortest path lengths were
listed and the largest one is 13.31, which is much more reasonable.

Could you please tell me how the shortest path length and the suboptimal
path length are computed in the NetworkView plugin?

Thanks so much for your help!

Best,
Yuchen