Table 1.
The statistics of experimentally verified S-nitrosylation sites in training set and independent test set.
Figure 1.
The flowchart of MDD clustering.
Figure 2.
Frequency plot of sequence logo of S-nitrosylation sites with 21-mer window length.
Figure 3.
The compositional biases of amino acids around S-nitrosylation sites compared to the non-S-nitrosylation sites.
The amino acids that are significantly enriched or depleted (P-value<0.05) around S-nitrosylation sites are presented.
Figure 4.
Comparison of average percentage of ASA in the 21-mer window (−10∼+10) between S-nitrosylation and non-S-nitrosylation sites.
Table 2.
The cross-validation performance of the models trained with various features.
Figure 5.
The top twenty physicochemical properties of S-nitrosylation sites ranked by the average value of F-score measurement in 21-mer window.
KRIW710101, side chain interaction parameter [48]; OOBM850105, optimized side chain interaction parameter [49]; FINA910104, helix termination parameter at position j+1 [50]; FAUJ880111, positive charge [51]; GUYH850104, apparent partition energies calculated from Janin index [52]; KIDA850101, hydrophobicity-related index [53]; GUYH850101, partition energy [52]; JANJ780101, average accessible surface area [54]; ROSM880102, side chain hydropathy [55]; CHOC760102, residue accessible surface area in folded protein [56]; FASG890101, hydrophobicity index [57]; RACS820103, average relative fractional occurrence in AL(i) [58]; KARP850103, flexibility parameter for two rigid neighbors [59]; OOBM770101, average non-bonded energy per atom [60]; LEVM760101, hydrophobic parameter [61]; MEIH800102, average reduced distance for side chain [62]; GUYH850105, apparent partition energies calculated from Chothia index [52]; KRIW790102, fraction of site occupied by water [63]; PUNT030101, knowledge-based membrane-propensity scale from 1D_Helix in MPtopo databases [64]; WEBA780101, RF value in high salt chromatography [65].
Table 3.
The cross-validation performance of the models trained individually with twenty physicochemical properties.
Table 4.
The predictive performance of MDD-clustered models using an independent test set (GPS-SNO).
Figure 6.
A case study of Bos taurus dimethylarginine dimethylaminohydrolase 1 (DDAH1) which contains two S-nitrosylation sites at positions 222 and 274.