This gzipped tar archive contains the source required to compile an image classification program based on the work of Ogura et al (see references). It is written in C++ and includes a README file that covers its operation.

The program can also be used in the generation of classes for Random Conical Tilt or Orthogonal Tilt Reconstruction using SPIDER.

Adapt is a piece of working, in-house software and as such hasn’t been extensively tested on many hardware setups and systems.

Note: Release 2.02 compiles cleanly under gcc 4.3.x and includes a bugfix for a segfault condition in the previous releasee.


Required Inputs

  • an IMAGIC image stack, most likely the output of a multi-re2ference alignment (MRA), to be classified into groups.
  • various parameters of classification described in the README and Ogura paper.
  • the number of nodes in the network (e.g., the number of classes wanted).


  • an IMAGIC image stack corresponding to ‘node images’ (see Ogura paper)
  • an IMAGIC image stack of class averages
  • SPIDER format select files for each class.


Ramey VH, Wang HW, Nogales E. (2009) Ab initio reconstruction of helical samples with heterogeneity, disorder and coexisting symmetries. J Struct Biol., E-pub ahead of print.

Ogura T, Iwasaki K, Sato C. (2003) Topology representing network enables highly accurate classification of protein images taken by cryo electron-microscope without masking. J Struct Biol.143(3), 185-200.


Download Adapt2_02_mp.tar.gz (12k)