File formats relating to structure

BMERC : needle tools : File formats : Structure file formats

This includes secondary structure as well as tertiary structure. The former is usually based on the DSSP program, and the latter on PDB ATOM format.

Table of contents

1. File formats relating to structure
   1. Table of contents
   2. Segment file format
   3. Abbreviated DSSP file format
   4. .nexp (exposure) file format
   5. Core file format
   6. Internal "ss" format
   7. Internal "core abstraction" file format

1. chain ID ( chain ID; one character) -- identifies the chain for this segment. This must always be exactly one character; for many of the single-chain PDB entries, it will consist of a space character (" ").
2. index (integer) -- segment index. These are as assigned in the dssp4.pl program's -ss output; since this output is generally not saved, these numbers are therefore not useful except in debugging.
3. starting residue (pdbres) -- first inclusive residue of the segment.
4. ending residue (pdbres) -- first inclusive residue of the segment.
5. structure type (segment designator) -- alphanumeric segment type designator, e.g. "H", "E2", or (for barrels) "C030204".
6. segment length (integer) -- length in residues. (This field was added in Release 1.0.)

See [Kabsch,W., and Sander,C., Biopolymers 22(1983) 2577-2637], or the DSSP web pages for more information on DSSP, and how to obtain a copy. See the Description of the DSSP program page for details of how to run dssp and how to interpret the "raw" DSSP output format.

An abbreviated DSSP file has a single header line, followed by a series of records each with eight tab-delimited fields. There is one record per residue, plus extra records with an "amino acid" of "!" to denote chain breaks or transitions between chains. Some of these fields, described below, have subfields.

1. the "pdbres" field followed by the chain ID (6 characters total). This serves to completely identify the residue within the PDB file, though normally the chain ID appears immediately before the pdbres (residue index and insertion code) field, as in the PDB ATOM records.
2. the one-letter amino acid abbreviation (1 character).
3. the DSSP structure designation codes (9 characters). See the "Output" section of the DSSP: Program description page for details of how to interpret these codes.
4. the "BP1" field (4 digits) and
5. "BP2" field (4 digits followed by 2 alphas), which identify the "bridge partner" to which a strand residue is hydrogen bonded. The numbers are 1-based residue indices which include chain breaks and encompass the whole file, allowing inter-chain hydrogen bonding to be described. To find the bridge partner residue in emacs, type "M-x goto-line RET" and then a number one greater than the one in the appropriate field -- one greater because of the header line.
6. normalized solvent accessibility [explain. -- rgr, 28-Jan-97.]
7. backbone phi and
8. psi angles.

DSSP state codes:

These are interpreted in the order given; if more than one applies, the first is chosen. (Based on [Kabsch&Sander].)

Letter	Name	Definition
H	Alpha helix (4-12)	Two or more consecutive bridge partners at i and i+4.
B	Isolated beta-bridge residue	Must not have a neighbor that qualifies it for H, E, G, or I status. Bridge partner is identified in BP1 or BP2 column.
E	Strand ("extended")	Has at least one bridge partner and at least one neighbor bridged in parallel or antiparallel.
G	3-10 helix	Two or more consecutive bridge partners at i and i+3.
I	pi helix	Two or more consecutive bridge partners at i and i+5.
T	Turn	Bridge partner at i+3, i+4, or i+5, but no bridged neighbor that would qualify them for H, G, or I status.
S	Bend	Local curvature greater than 70 degrees, measured as the angle between alpha carbons at i-2, i, and i+2.
blank	None	Meets none of the criteria above.

Backbone phi and psi angles, chirality, disulphide bonds, and solvent exposure are also computed, but do not affect the state code.

Known bugs:

1. There is no room in this format to accomodate NMR structures. These typically use the PDB MODEL/ENDMDL construct to delimit two or more variations on the same structure, sometimes as many as 50 of them. [Should discuss this more fully under the generate-dssp program.] -- rgr, 28-Apr-97.

1. residue index (columns 1-5) -- the PDB sequence number and insertion code for the residue (blanks are significant).
2. AA code (column 7) -- the standard one-letter amino acid abbreviation.
3. chain ID (column 9) -- the PDB chain identifier.
4. exposure value (columns 44-50) -- exposure value, usually in Fortran F7.3 format.

Old files may have more fields, not documented here; the extra whitespace is for compatibility. Note that old files will NOT have the chain ID, though the space that will be in that position should serve as such.

1. A secondary structure designator, which occupies the entire line and denotes the start of a new core segment; and
2. A series of PDB-format ATOM records that describe the backbone of the core segment, plus beta carbons (including beta carbons "hallucinated" for glycine residues).

Multiple chains may appear in a core file, but it is assumed that it makes sense to thread the entire model with a single sequence in the order in which the segments appear.

Residues within each core segment and the segments themselves within the file appear in amino to carboxyl order. The backbone must be contiguous, and each residue will have exactly five atoms that appear in the order N, CA, C, O, and CB. It is recommended that core files that are not generated automatically by make-core.pl be passed through filter-pdb-atoms.pl in order to guarantee these conditions. Use

 
	filter-pdb-atoms.pl -output cb -hcb -pass-through all \
		handmade.core > checked.core

1. the chain ID.
2. the "pdbres" field.
3. the 1-char amino acid abbreviation.
4. the 1-based secondary structure index, assigned in chain order, with all chains and all secondary structures (both helices and strands) on the same numbering system).
5. the first character of the smoothed DSSP secondary structure assignment (H or E).
6. the residue indices of H-bonded strands, when available, else "" or 0 if not (two values with a tab in between). These indices are 1-based, and count all residues in all chains, including the "!" entries for chain breaks and terminations. (0 means a B or short E strand that was eliminated in the smoothing process.)

A core abstraction file is a series of lines, each of which contains one or more tab-delimited fields, with the first field being the record tag. All tags and string values are in lower case.

seg type length x1 y1 z1 x2 y2 z2

Starts a new segment, where

• type is a segment type designator as from the original core file,
• length denotes the number of residues;
• x1 y1 z1 gives the coordinates of the segment's first residue's backbone N atom; and
• x2 y2 z2 gives the coordinates of the segment's last residue's backbone C atom.

After the seg record, there must be exactly length res records before the next seg or loop record.

res exposure [ vv-exposure ]

Denotes the exposure for the residue. exposure may be the empty string if no EFA exposure was specified when the file was created. If no visible volume "exposure" was specified, then there will be no vv-exposure field.

loop loop-type loop-length

Describes a loop. Loops may appear at the beginning of the file (in which case the loop-type is "a" for the amino leader), after the last res record of one segment but before the next seg (in which case the loop-type is "i" for "internal"), or at the end of the file (in which case the loop-type is "c" for the carboxy trailer). Loop lengths of 0 are possible. Loop records are entirely optional (but all should be present if any are included).

As an example, here is the entire contents of 1puc.cab, a core of minimal (if not sub-minimum) size:

loop	a	10
seg	H	5	13.361	13.423	57.722	21.622	13.524	57.136
res	121.997
res	103.892
res	21.241
res	41.551
res	81.506
loop	i	51
seg	H	6	11.361	27.337	46.732	16.827	21.339	52.382
res	72.848
res	80.841
res	1.468
res	15.439
res	66.237
res	29.616
loop	c	33