I want to try if RoseTTAFold can well predict the protein complex such as virus-host interactions. So I used the protein sequences of SARS-CoV-2 RBD and human ACE2 as input following the README instruction in RoseTTAFold-main\example\complex_modeling\README.
I tried to make "good" paired sequence alignments to extract coevolution signal and also filtered the paired alignment using hhfilter(command as this : hhfilter -i SARS2-Spike-RBD_7DF4_1-ACE2.a3m -o processed_SARS2-Spike-RBD_7DF4_1-ACE2.a3m -id 97 -cov 97). Then ran complex structure prediction with this command : predict_complex.py -i processed_SARS2-Spike-RBD_7DF4_1-ACE2.a3m -o SARS2-Spike-RBD_7DF4_1-ACE2_complex -Ls 223 616
Finally, i got the pdb files of RBD-ACE2 complex. But the result is so strange and very different from the actual structure i downloaded from PDB database. Almost all the structure are bold loops with no α-helix and β-sheet existing.
[Uploading Complex modeling.zip.004.zip…]()
The sequence files i used and the results of PDB files and png files are attached. Can someone help me with the problem why the PDB outcome is so strange and far from the actual structure?
Complex modeling.zip
Dear all,
I want to try if RoseTTAFold can well predict the protein complex such as virus-host interactions. So I used the protein sequences of SARS-CoV-2 RBD and human ACE2 as input following the README instruction in RoseTTAFold-main\example\complex_modeling\README.
I tried to make "good" paired sequence alignments to extract coevolution signal and also filtered the paired alignment using hhfilter(command as this : hhfilter -i SARS2-Spike-RBD_7DF4_1-ACE2.a3m -o processed_SARS2-Spike-RBD_7DF4_1-ACE2.a3m -id 97 -cov 97). Then ran complex structure prediction with this command : predict_complex.py -i processed_SARS2-Spike-RBD_7DF4_1-ACE2.a3m -o SARS2-Spike-RBD_7DF4_1-ACE2_complex -Ls 223 616
Finally, i got the pdb files of RBD-ACE2 complex. But the result is so strange and very different from the actual structure i downloaded from PDB database. Almost all the structure are bold loops with no α-helix and β-sheet existing. [Uploading Complex modeling.zip.004.zip…]()
The sequence files i used and the results of PDB files and png files are attached. Can someone help me with the problem why the PDB outcome is so strange and far from the actual structure? Complex modeling.zip