Community rating: unrated.
GOSeq is based upon Young et al, Genome Biology 2010, 11:R14 "Gene Ontology analysis for RNA-Seq: accounting for selection bias" (http://genomebiology.com/2010/11/12/220).
This tool does a GO enrichment analysis from a differentially expressed list of transcripts, correcting for length biases inherent in RNASeq. Note that standard GO enrichment analysis tools (e.g. EasyGO, GSEA, DAVID, etc) assume that all genes are independent and equally likely to be selected as differentially expressed; in RNASeq long transcripts and highly expressed transcripts are more likely to be over-represented.
GOseq is also available as a package from bioconductor, provided by WEHI. The latest release as well as up to date documentation can be obtained directly from bioconductor package (http://www.bioconductor.org/packages/release/bioc/html/goseq.html).
The input file is a tab-separated list of transcript IDs, differential expression status (0|1), transcript length and GO:IDs. Multiple GO:IDs are pipes separated (|). Differential expression status is indicated by either 0 or 1 where 0 means not differentially expressed, 1 is differentially expressed.
For example, a line in the input file will look like:
ENSG00000183798 0 3831 GO:0005575|GO:0005576|GO:0005578
Where there is no GO annotated to the transcript, the GO field can be left blank.