Construction of AMPs profiles (text mining)

The Hidden Markov Models (HMMER) algorithm version 2.3.2 [52] was used to construct specific pathogen-targeted models/profiles using the respective training datasets. All the HMMER profiles were built on the Ubuntu 12.04 LTS operating system. The task was accomplished on a terminal and the command lines used to build each profile was written following the corresponding algorithm and the steps involved in their construction were as below:

For the first step, the training datasets of each target class were aligned using the ClustalW alignment tool [53]. The alignment was carried out using the command line:

The command line simply states <<do an alignment of the sequences which are in the upper case found in the input file “target class.fasta” with the FastA, using ClustalW as multiple alignment tools and GCG Postscript output for graphical printing>>. The output of the command results in the construction of aligned sequences, called “target class.msf”. The aligned sequences were used as input in the next step.

The next step enhances the construction of the profiles of the target class sequences by showing the common motifs/signatures within the profiles. To achieve this, the “Build profiles” was run using the following command:

To enhance the sensitivity of the profiles, the file generated (target class. hmm) from the profile building step was calibrated by using the command line:

The resulting profiles “target class.hmm” was used in evaluating the performance of the profiles by testing the created profiles on an independent AMP dataset.