Python Programming for Biology: Bioinformatics and Beyond

Determining macromolecular structures

There  are  a  variety  of  methods  that  can  be  used  to  determine  the  three-dimensional

structures  of  large  biological  molecules,  to  various  degrees  of  precision  or  resolution.

However,  the  vast  majority  of  high-resolution  structures  of  biological  molecules  were

determined  by  just  two  experimental  techniques:  X-ray  crystallography  and  nuclear

magnetic resonance.

X-ray  crystallography,  as  you  might  guess,  works  by  forming  crystals  made  of  the

molecule  of  interest,  and  is  the  most  common  form  of  macromolecular  structure

determination.  Growing  the  crystals  to  start  with  is  often  the  most  difficult  part  of  the

process.  However,  if  good  crystals  can  be  made,  then  because  the  molecules  are  held

together  in  a  lattice,  the  regular  array  of  atom  positions  acts  to  diffract  a  beam  of  X-ray

radiation,

5

 thus  creating  a  diffraction  pattern  consisting  of  spots  called  reflections,  once

the  beam  has  passed  though  the  crystal.  Several  diffraction  patterns  are  collected  by

rotating  the  sample  in  a  precisely  known  way,  to  view  the  structure  from  a  variety  of

angles. The diffraction patterns can be combined, by mathematical means that we will not

go  into,  to  derive  a  three-dimensional  map  of  the  electron  density  of  the  atoms.  The

---

structure  of  the  macromolecule  is  then  determined  by  fitting  its  covalent  chemical

structure into the electron density to find the best fit. For proteins this usually involves a

computer  program  that  first  finds  the  backbone  route  (fairly  easy  because  of  the

characteristic  spacing  and  angles  along  the  polypeptide)  and  then  fits  the  side  chains,

although this last part often needs manual assistance.