Methods and data structures are disclosed for carrying out identifying differences between large files comprising many lines (or other units of comparison such as rows, words, paragraphs, sentences, etc.). The disclosed methods and data structures facilitate and carry out a streamlined, yet thorough comparison of two files to identify differences between them. The streamlining is achieved by pre-processing the files prior to submitting them to any known longest common subsequence (LCS) search engine. The output of the LCS generator is post-processed to compensate for changes to the sequences introduced by the pre-processing stage.