To score abstracts based on the frequencies of discriminating words: 1. score_abstracts.pl Abstract_file Discriminating_Word_file > output_file Abstract_file should be in following format: Abstract_PMID_1 Text of abstract on same line, no carriage returns Abstract_PMID_2 Text of next abstract... .... etc... Discriminating Word file should be in following format: ACTIN n: 188 f: 4.2e-03 bk_f: 2.4e-04 p: -364.13 ... etc... where only the word, frequency in positive abstracts (value following "f:"), and frequency in negative abstracts (value following "bk_f:") are actually used by this program. 2. sort output_file according to score, e.g. by using UNIX sort command: sort -n -k 4 output_file > sorted_output_file