DFLpred is designed for the prediction of disordered flexible linker (DFL). The method generates numeric score for each residue in the input protein sequence that quantifies putative propensity to form a DFL. Larger value of propensity denotes higher likelihood to form DFL. It also provides putative binary annotations (a given residue is predicted either as a DFL or not a DFL) based on false positive rate = 0.05 using threshold on the propensity score = 0.18). Residues with propensity > 0.18 are assumed to form DFLs and otherwise they are assumed not to form DFLs.
Java Runtime Environment 1.5 or later
Any ANSI C compiler e.g. GUN C compiler to compile "myiupred.c" (if "myiupred" is not runnable on your computer)
java -jar DFLpred.jar [input_fasta_file] [output_file]
Accepts up to 5000 FASTA sequences in a single fasta file
java -jar DFLpred.jar examples.fasta results.txt
Line 1: >protein ID
Line 2: protein sequence using 1-letter amino acid encoding. Lower/upper case indicates a residue is/is not predicted as a flexible disordered linker (DFL) by the score threshold 0.18
Line 3: comma-separated propensity score of each residue to be a DFL
http://biomine-ws.ece.ualberta.ca/DFLpred
Meng, F. and Kurgan, L. DFLpred: High-throughput prediction of disordered flexible linker regions in protein sequences. Bioinformatics 2016;32(12):i341-i350
We acknowledge with thanks the following software used as a part of DFLpred:
IUPred (http://iupred.enzim.hu/) - Prediction of Intrinsically Unstructured Proteins
AAindex (http://www.genome.jp/aaindex/) - Amino acid indices, substitution matrices and pair-wise contact potentials