Date: Mon, 03 Sep 2007 15:14:53 -0500 From: Yang Zhang Subject: Re: I-TASSER energy potential questions Hi, >1) What are the distance "bins" used for r13, r14 and r15 potentials? > >There are 2 values for r13. Do they correspond to bins: < 4A, > 4A >There are 3*8=24 values for r14. Do the correspond to bins in >range(-12A,12A) with step 1A? The r14 potential also seems to be >chirality depended (probably half is used for the left, half for the >right-handed conformations). >They are 2*8=16 values for r15 . Do they correspond to bins >in range(1A,8A) with step 0.5A? For r13.comm, the two values correspond the potential of r(i,i+2)<6.03A, >6.03A. For r14*.comm, you are correct. The first half is left-hand and second right-hand. For r15*.comm, it is for the range (1A,16A) with step 1A. >2) How is the final potential for r14 and r15 decided? > >I know it is a combination of general (r1x) and secondary structure >specific potential (r1xe/r1xh). Is this just an arithmetic mean? If the local secondary structure is helix (i,i+1,i+2,i+3 are helix), we use r14h.comm; if the local secondary structure is strand (i,i+1,i+2,i+3 are strand), we use r14e.comm; otherwise we use r14.comm. We found it works better than averaging r14.comm and r14h.comm. r15*.comm is similar. >3) Are the short range potentials E'12 to E'15 for side chain >interactions from TOUCHSTONE-II article used by I-TASSER? > >As TASSER articles mention modification in the short range force field >I guess that since you did not list them on the web site, they might be >not used anymore to calculate E_short. Is that correct? Yes, we do not use these terms anymore. >4) How are the contact cutoff values related to R_min and R_max from >TOUCHSTONE-II force field? > >I guess in I-TASSER the fixed 3.25Å hard core and range(3.25Å,5Å) soft >core is not used anymore. The comments in potential file mention >addition of 2.5 standard deviation to the mean. Is that a soft core >range now with the mean being hard core diameter? The implementation of hard/soft core in I-TASSER is different. The R_min and R_max are decided by the values (R0) in the last 3 matrix in 'quasi3.comm', i.e. R_min=(1-0.17)R0; R_max=(1+0.17)R0. In 'contact.comm', the cut+2.5 standard deviation is for the cutoff used for external contact restraints (e.g. threading-based contact predictions). >5) What is the meaning of two groups of quasi (pairwise >orientation-dependent) potentials? > >The 2D matrix of values is given for each pair of amino acids and >parallel, anti-parallel and perpendicular orientations. However, >instead of 3 matrices there are 6 of them. What is the meaning of >this two groups? Is it dependency on predicted secondary structure? The first three are the potential (E0), the last three are the cutoff (R0). In real implementation, we have made E0=E0-0.25 for |i-j|>5 to encourage long-range contacts. >6) How to interpret the profile potential? > >For each amino acid 9 matrices of 9x9 values are given, so it looks as >the 3D matrix for parallel, anti-parallel and perpendicular residues in >contact. What "bins" are used here? Is it just the range(0,8)? The following is the code for reading the table. The contact cutoff is soft core cutoff R_max as mentioned above. c PROFILE3 potential =envir(#of antiparalel contacts, c #of orthogonal, # of parallel, aminoacid's type, ith residue) do i=0,19 !from column read(2,*) do im=0,8 do ia=0,8 read(2,*)(envir(ia,im,ip,i),ip=0,8) end do read(2,*) end do read(2,*) enddo Hope this will help. Yang