patGPCR: A Multitemplate Approach for Improving 3D Structure Prediction of Transmembrane Helices of G-Protein-Coupled Receptors

Algorithm 1

Refinement by single pipeline (𝑃,𝑇𝑀[], 𝐿𝑜𝑜𝑝[]).

(1) Input:  𝑃  is starting conformation, TM [] is the set of residue numbers for  TM

    regions, Loop [] is the set of residue numbers for loop regions.

    Output: conformation after  TM  refined

    𝐸_𝑚  is the Rosetta membrane energy function with S.

    𝐸_{𝑠𝑐𝑜𝑟𝑒3} is the Rosetta score3 energy function.

(2)  𝑑^→_𝑇𝑀[] = getHelixAxis();

(3)  𝑑^→_𝑀 = getMembrane();

(4)  for  𝑖=1 to STAGE2_CYCLES  do

(5) 𝑃^′ = transferHelix(P, TM [], RND(−5, 5), RND(−5, 5), RND(−5, 5));

(6) if (𝐸_𝑚(𝑃^′)<𝐸_𝑚(𝑃)) 𝑃←𝑃^′;

(7)  end for

(8)  for  𝑖=−180 to 180  do

(9)     𝑃^′ = spinHelix(P, TM [], 𝑑^→_𝑇𝑀[], 𝑖);

(10) if (𝐸_𝑚(𝑃^′)<𝐸_𝑚(𝑃)) 𝑃←𝑃^′;

(11)  end for

(12)  for  𝑖=1 to STAGE4_CYCLES  do

(13) 𝑃^′ = tiltGaussianHelix(P, TM [], 𝑑^→_𝑀, 30, 6);

(14) if (𝐸_𝑚(𝑃^′)<𝐸_𝑚(𝑃)) 𝑃← 𝑃^′;

(15)  end for

(16)  LoopModelRandomly(Loop [], 𝐸_{𝑠𝑐𝑜𝑟𝑒3})

(17)  P = crossHelices(P, TM []);

(18)  Output: 𝑃;