1. Question: How much memory and storage space does RAPTOR need? How long does it take to install?
2. Question: How long will it take to run one sequence?
3. Question: How does RAPTOR perform 3D modeling?
   
4. Question: Should I use NR instead of REFSEQ and how often should I update the database?
   
5. Question: Is it possible to integrate RAPTOR into our own pipeline?
   
6. Question: Is it possible to create our own template library?
   
7. Question: What is in the template files?
   
8. Question: How the templates are generated from the PDB?
   
9. Question: What are the pdb files for?
   
10. Question: Can RAPTOR predict structure for proteins with more than one domain?
    
11. Question: Can RAPTOR predict structure for trans-membrane proteins?
    
12. Question: What is minimum and maximum size of sequence that can be accepted by RAPTOR?
    
13. Question: Can RAPTOR be run on a cluster?
    
14. Question: Does RAPTOR support MPI?
    
15. Question: Can I use RAPTOR to do function annotation?
    
16. Question: Will RAPTOR's threading algorithm be further improved in the future?
    
17. Question: How do you compare RAPTOR with other similar software tools?
    
18. Question: Do you provide customization?
    

Question 1: How much memory and storage space does RAPTOR need? How long does it take to install?

Answer: A large memory, preferably 1GB, and multiple CPUs (AMD or Intel) are preferred for best performance. The installation needs about 3 GB of disk space, and may take up to 30 minutes to finish, depending on your hardware setup. The minimum configuration is 512Mb of RAM, 3GB free hard drive space and a CPU no less than a 1Ghz.

Question 2: How long will it take to run one sequence?

Answer: The run time is dependent on the sequence length. The longer the sequence, the longer the run time. Typically, for NoCore and NPCore methods, it will take about one hour to complete a sequence. For IP, the running time may be several hours.

Question 3: How does RAPTOR perform 3D modeling?

Answer: After threading, RAPTOR outputs the alignments and then calls Modeller to generate 3D structures. Modeller is free for academic users. After threading, RAPTOR can also generate all the input files required by ICM PRO from the alignments. If you have Modeller or ICM PRO, you can use them seamlessly with RAPTOR.

Question 4: Should I use NR instead of REFSEQ and how often should I update the database?

Answer: Due to the installer's size constraint, REFSEQ database is included in RAPTOR instead of NR database. The version of REFSEQ database (required by PSI-BLAST) included in RAPTOR may not be the most current one. You are encouraged to download the latest release from NCBI's ftp site (refer to RAPTOR manual about how to update REFSEQ). We have conducted a comparison between REFSEQ and NR. There is not a significant difference between the sequence profiles obtained from the two and the size of REFSEQ is just half that of NR. For best performance, you can download NR.01 and NR.02. (Refer to the RAPTOR user manual about how to install NR databases.)

Question 5: Is it possible to integrate RAPTOR into our own pipeline?

Answer: Yes, RAPTOR can be used on command line and you can obtain all the output from the XML output.

Question 6: Is it possible to create our own template library?

Answer: Yes, you can. However, you will need to contact BSI if you have such requirement for assistance.

Question 7: What is in the template files?

Answer: The template files are of two types: .psm stores the position specific matrix of the template. and .fssp contains the coordinates of the backbone atoms of the template. They are used in threading.

Question 8: How the templates are generated from the PDB?

Answer: First, we get all the sequences whose structures are in the PDB. Then we use some software to generate PDB40 sequence set. Every two sequences in PDB40 has a maximum sequence identity of 40%.

Question 9: What are the pdb files for?

Answer: The pdb files are for 3D modeling. To save disk space, the pdb files have been trimmed.

Question 10: Can RAPTOR predict structure for proteins with more than one domain?

Answer: For best performance, you may need to cut a multiple-domain protein into domains first and run each domain with RAPTOR.

Question 11: Can RAPTOR predict structure for trans-membrane proteins?

Answer: No, RAPTOR cannot do that.

Question 12: What is minimum and maximum size of sequence that can be accepted by RAPTOR?

Answer: From 25 to 1500 AA.

Question 13: Can RAPTOR be run on a cluster?

Answer: Yes, RAPTOR can be run on a cluster.

Question 14: Does RAPTOR support MPI?

Answer: Yes.

Question 15: Can I use RAPTOR to do function annotation?

Answer: Yes. Proteins with similar 3D structures have similar functions. So the function of a query sequence can be predicted from the function of the top template.

Question 16: Will RAPTOR's threading algorithm be further improved in the future?

Answer: Yes. We have several projects ongoing which will create the next generation threading algorithm.

Question 17: How do you compare RAPTOR with other similar software tools?

Answer: Most other software are based on dynamic program and simple homology search. They can not produce globally optimum alignment if the scoring function considers pair wise interaction. But RAPTOR can. RAPTOR's integer programming techniques treat pair wise interaction rigorously.

Question 18: Do you provide customization?

Answer: Yes. We have customized RAPTOR for several pharmaceutical customers.