DeepFrag-K: Deep learning methods to directly classify any protein into one of the well-known folds.

Run : 1-Put the target protein fasta sequence in 'fasta.fa' file. the fasta sequence must be on the following format :

d1a4pa_ psqmehametmmftfhkfagdkgyltkedlrvlmekefpgflenqkdplavdkimkdldqcrdgkvgfqsffsliagltiacndyfvvhmkq 2- on the framework main directory using command line 'python deepfrag.py fasta.fa' to run the deepfragk prediector.

Output :

• Fold A.39
• FASTA PSQMEHAMETMMFTFHKFAGDKGYLTKEDLRVLMEKEFPGFLENQKDPLAVDKIMKDLDQCRDGKVGFQSFFSLIAGLTIACNDYFVVHMKQ
• FRAGMENTS FREQUANCY 10,4,5,10,8,4,1,8,7,11,0,3,1,6,6,2,4,4,11,8,8,3,0,1,3,1,8,9,25,2,7,5,4,10,16,12,1,29,6,4,7,2,10,7,5,0,7,9,3,7,2,11,1,9,5,4,9,8,3,9 ,0,2,1,7,7,5,9,2,7,5,8,3,5,6,2,3,5,6,0,8,7,2,7,15,0,28,9,0,1,13,9,4,16,5,8,5,4,3,22,11

Required Software Installation : Python2.6 MongoDB ubuntu server 18.01 LTS

Minimum Machine specx : 200 GB RAM MultiCore Processor (Tested on 40 Cores).

Python Packages Dependancies : pymongo sklearn 2.x.x pywt tensorflow keras shutil numpy pprint

Database: due to the size of the database, we provide a download link on request. please contact welhefna@odu.edu to provide the database access link.

Directories and Files: | |---fasta, fasta sequence processing, fasta sequence reader. |---pssm, position specific scoring matrix reader. |---vi, features calculations and scripts. |---deepfrag.py, main program. |---model.py, model reader and utilities. |---README.txt, description of framework.