cmap appears for force.renumber atoms

[Ron Ayoub]

I would like to create many pairs of contacts for many pdb files and I would like those indices in the cmap matrix to correspond to the resno fields in the atom records. As it stands, it appears the cmap forces a renumbering of the indices so that they start at 1. I'm not sure if it also compresses in the case of gaps. Currently, I have to force a renumbering of the pdb via trimming and then cleaning with the force.renumber options. Is there a better way to approach this?

[Xinqiu Yao]

Hi,

cmap() function takes care of gaps but ignores the resno field. Actually, I don't think it is a good idea to build a cmap matrix matching 'resno'. In many pdb files, 'resno' is not a unique identifier to distinguish residues. For example, residues in different chains can have the same resno. Residues can even have same chain ID and resno (with different 'insert' codes).

I suggest name rows/columns of the cmap matrix by corresponding resno (or a combination of chain ID, resno, and insert code). Then, you can map residues to contact map without renumbering the pdb.

Let me know if it works for you.

[Ron Ayoub]

Thanks for the response.

For the discussion below, assume I ignore inserts and have already handled multiple chains.

I'm interested in extracting contact differentials between residues. I have selected only carbon alpha atoms and hence have only 1 atom per residue. The distance between residues as numbered is important to maintain in the case of missing residues. For instance, if resno 4 contacts resno 8, even if resno 6 and 7 are absent, that differential should still be maintained as 8 - 4 = 4 since those intervening residues do exists in the chain whether or not their position was discovered by crystallography.

After melting the cmap matrix I store a data frame called cmap.pairs. I just convert the serial indices in the cmap to the actual resnos obtained from the selection of carbon alphas. above.

    cmap.pairs$Var1 <- cmap.props$resno[cmap.pairs$Var1]
    cmap.pairs$Var2 <- cmap.props$resno[cmap.pairs$Var2]

I think this is sufficient for now. Inserts are an issue and occasionally I see an old pdb where the resnos are not consecutive and not because of missing residues. I have not determined how often that happens and if I should worry about that yet.

I do think that the way I solved my issue may be the prescriptive way to solve it using bio3d and I'm not sure it is the responsibility of bio3d to handle this.

Thanks again for a great tool to work with.