Abstract
BACKGROUND: Immunoglobulin rearrangement involves random and imprecise processes that act to both create and constrain diversity. Two such processes are the loss of nucleotides through the action of unknown exonuclease(s) and the addition of P nucleotides. The study of such processes has been compromised by difficulties in reliably aligning immunoglobulin genes and in the partitioning of nucleotides between segment ends, and between N and P nucleotides. RESULTS: A dataset of 294 human IgM sequences was created and partitioned with the aid of a probabilistic model. Non-random removal of nucleotides is seen between the three IGH gene types with the IGHV gene averaging removals of 1.2 nucleotides compared to 4.7 for the other gene ends (p < 0.001). Individual IGHV, IGHD and IGHJ gene subgroups also display statistical differences in the level of nucleotide loss. For example, within the IGHJ group, IGHJ3 has average removals of 1.3 nucleotides compared to 6.4 nucleotides for IGHJ6 genes (p < 0.002). Analysis of putative P nucleotides within the IgM and pooled datasets revealed only a single putative P nucleotide motif (GTT at the 3' D-REGION end) to occur at a frequency significantly higher then would be expected from random N nucleotide addition. CONCLUSIONS: The loss of nucleotides due to the action of exonucleases is not random, but is influenced by the nucleotide composition of the genes. P nucleotides do not make a significant contribution to diversity of immunoglobulin sequences. Although palindromic sequences are present in 10% of immunologlobulin rearrangements, most of the 'palindromic' nucleotides are likely to have been inserted into the junction during the process of N nucleotide addition. P nucleotides can only be stated with confidence to contribute to diversity of less than 1% of sequences. Any attempt to identify P nucleotides in immunoglobulins is therefore likely to introduce errors into the partitioning of such sequences.