structure information is added). When the method is ran
without structural information, the underlying algorithm
that complete the prediction task is ConSeq [22]. For this
example, use default parameters for homolog search. It is
possible to enter a job title and an e-mail address to get the
link of the results (optional). The number of homologous
sequences is a limitation for the Consurf run (seeNote 6).
Clicking “Submit” button will start the analysis, which can
take from minutes to a few hours according to the query.
Submitting the job will open the job status page that
shows running parameters, run process, and running mes-
sages. When the job is completed, the results page will include
links for the final prediction results, input sequence data, the
MSA, phylogenetic tree, and additional links for high resolu-
tion figures including the protein structure. In the output, the
first option displays the conserved sites/regions, colored on
the sequence according to the conservation scores. In Con-
Surf a conservation scale from 1 to 9 was employed (with
distinct colors), where 1 represents the least and 9 represents
the most conserved residues. Additionally, ConSurf predicts if
the residues are buried or exposed to the solvent. Predictions
for functional residues (conserved and exposed) and structural
residues (conserved and buried) are also labelled with different
colors below the sequence (seeNote 4). It is important to note
that the sequence based output is provided only when the
structural information is not included at the input level. If
the structure information is added in the first place, the con-
servation predictions are shown as marked on the 3D struc-
ture. Furthermore, ConSurf produces a phylogenetic tree
using the MSA output. It is possible to download the whole
output as a zip file.(c) Machine learning based approach (PROFisis)
Run the program forPQvia the PredictProtein (https://
http://www.predictprotein.org/)) service, which normally requires
registration (seeNote 7). Enter the query amino acid sequence
and hit the PredictProtein button. When the results are gen-
erated, it will be displayed automatically on the screen. In the
output, predicted binding sites for protein interactions, sec-
ondary structure, solvent accessibility, transmembrane helices,
disordered and flexible region, disulfide bridges, subcellular
localization, gene ontology terms of proteins, and effect of
point mutations are shown. Results can be displayed separately
by clicking relevant part from the left panel or all-in-one by
clicking the “Dashboard.” In the figure, place the cursor over
the interested region to observe the annotation details includ-
ing the annotated residues, sequence length, type of the pre-
diction and its evidence, for the selected region. It is also
Phylogenetics-Based Prediction of Functional Sites 63