Because of the number and independence of the programs in the package it is very desirable that the input format for all programs be the same. Although some data file are created by the computer and others by the user, the style of input has been designed with emphasis on the benefit of the user. All input is token based. The input is read in free format, each token being separated from the next by a delimiter such as a space or a comma (see Tables 1 and 2 for examples).
Table: Definition of the general data cards
Table: Definition of the structure of Cro repressor
The first token on the card is the keyword. Keywords fall into two classes: data keywords and command keywords. The order of data cards is unimportant. When a command card is encountered, the required operation is performed on whatever data have been read to that point. Input and output is handled by a set of library routines which perform the basic operations of reading in cards, separating tokens, and building numbers from particular tokens.
A related objective was to allow the constituent programs to be easy to write and understand. Sets of common routines have been placed in libraries that are used by most of the programs. These libraries contain routines that process data cards and build the internal data structures. Also there are other routines that locate required information within the data structures. By not having to rewrite these routines one can implement a new program in a very short time and, by having a common internal structure, the existing programs can be understood more easily. On the other hand, although these library routines are available, they do not have to be used when one wishes to add a new program to the refinement package.