There are also a large number of avenin DNA sequences in Genbank, but not formally reported or analyzed in the literature, that have originated from both a variety of A. sativa cultivars and from other SB203580 species of Avena. These latter sequences were generated via PCR and represent an unknown coverage of the total avenin family. A recent report states, without details in the analysis or giving sequences, that there were eleven different avenin sequences found within the ESTs of a single oat cultivar. For the oat globulins, there has as yet been no complete oat globulin sequence family. Missing for both the oat avenins and globulins is such a detailed and comprehensive description. Although the best method of identifying complete gene sets is using a high-quality complete genome sequence, no such resource is currently available for oats, nor is likely to be available for some time. As a substitute, if sufficient high-quality ESTs are available, sets of active gene coding region sequences can be assembled. It is also best to use ESTs from single germplasms to avoid the complications of sorting out allelic sequences and differing gene set compositions. In the current report, ESTs for the oat cultivar CDC Dancer are used to assemble a proposed complete set of active oat avenin and globulin seed protein gene coding and derived amino acid sequences. Nine unique sequences are identified for the avenins and 24 for the globulins. The sequences of both classes generally agree with previously reported sequences. In addition, novel classes are reported. The composition and structure of the sequence families are analyzed and discussed along with evolutionary aspects of the families, relative representation of sequences in the EST resource, and issues in nomenclature within the grass prolamins. Individual contigs were examined for potential chimeric ESTs causing a truncation in contig assembly. Suspected chimeras were checked by BLAST analysis and ESTs with 59 or 39 sequence not matching avenins were removed. In cases where most of the EST sequence was UTR, the remaining coding sequence was too short for reliable assembling and such ESTs were also removed. The remaining ESTs were reassembled and the resulting contigs were divided into classes according to having five or more ESTs or fewer than five. The former were assembled together to find overlapping sequences not initially joined by the software. The resulting contigs were individually assembled with ESTs from the contigs with fewer than five ESTs and those matching exactly were merged into the larger contigs. Finally, each larger contig was manually inspected for ESTs apparently not matching the contig. These were removed and compared to the other contig consensus sequences and merged if found a match otherwise were left as unassigned. To check if ESTs might have been assigned incorrectly, possibly due to sequence similarity among avenin gene family members, the ESTs assigned to each final contig were assembled with the other eight contigs consensus sequences.