Detecting patterns of horizontal gene transfer (HGT) in genomic sequences is an important problem, with implications for evolution, ecology, biotechnology and medicine. Extensive genetic, biochemical and genomic studies have provided a good understanding of sequence features that are associated with many (though not all) known mobile elements and mechanisms of gene transfer. This information, however, is not currently incorporated into automated methods for gene transfer detection in genomic data. In this review, we argue that automated annotation of sequence features associated with gene transfer mechanisms could be used both to build more sensitive, mechanism-specific compositional models for the detection of some types of HGT in genomic data, and to ask new questions about the classes of genes most frequently transferred by each mechanism. We then summarize the genes and sequence features associated with different mechanisms of horizontal transfer, emphasizing those that are most useful for distinguishing types of transfer when examining genomic data, and noting those classes of transfers that cannot be distinguished in genomic data using existing techniques. Finally, we describe software, databases and algorithms for identifying particular classes of mobile elements, and outline prospects for better detection of HGT based on specific mechanisms of transfer.