Supplementary material for the manuscript "Improved success of phenotype prediction from the human immunodeficiency virus type 1 Env V3 region using neural networks"

Data set Format Comment
NSI/SI sequence set fasta format

V3 amino acid sequences in fasta format;
header contains sequence name, index numbers (ind), numbers identifying the patient of origin (pat), NSI/SI phenotype (phen), and a citation of the source of the sequence.
list format contains the same information as the fasta file but is suitable for importing into Excel
R5/X4 fasta format V3 amino acid sequences in fasta format;
header contains sequence name, index numbers (ind), numbers identifying the patient of origin (pat), R5/X4 phenotype (phen), and a citation of the source of the sequence.
list format contains the same information as the fasta file but is suitable for importing into Excel
Unphenotyped fasta format 1997 V3 amino acid sequences in fasta format; all sequences are epidemiologically unrelated.
name list list of names of the unphenotyped V3 sequences for retrieval from GenBank

Get the trained Neural Net

We've assembled a version of the trained neural network for HIV-1 subtype B coreceptor usage and MT-2 phenotype prediction. The archive files below contain the neural networks implemented in Matlab, a Python script for converting single-letter amino acid codes into a numerical encoding, and an example script demonstrating how the analysis was performed.

You'll need Matlab and the Matlab neural net toolbox to run this software.

NEUNET_README
Summary of archive contents
neunet.zip
A zip archive created with zipit.
neunet.sea
A self-extracting Mac archive created with zipit.
neunet.exe
A self-extracting PC archive created with zipit.