The visual SLR output for the three variants of the TH locus (AC132217.7). At the top, the horizontal dashed lines indicate which exon belongs to which transcript (from top to bottom, 001 in red, 002 in green and 003 in blue). The colour of the circles (SLR score) and bars (confidence intervals) denote selection mode. Below this is a per nucleotide measure of conservation with abnormally fast sites coloured orange (third codon positions) or red (first or second codon positions). The black hatching at the foot of the record of the number of sequences available at each alignment position. The whole of the second coding exon between positions 100 and 200 is clearly differently conserved, suggesting that it is under unusual selective pressures. This exon is present in variant 002 and 003. On the basis of the SLR output, we rejected the hypothesis that these two variants could give rise to the principal functional isoform.

Mapping functional residues. The output from firestar, showing the N-terminal end of the alignment between variant 001 of RAD50 (AC0040401.1, labelled as ‘Query’) and a sequence with known structure (1xexA). The top line indicates the residue number of the variant. Below the alignment the numbers in coloured squares indicate locally conserved regions. The last two lines indicate the presence of ligand binding residues for ATP and magnesium. These last two rows are colour-coded: the darker the colours, the more conserved the residue. The structure clearly has conserved ATP binding residues and these are unlikely to have arisen by chance. Variant 002 (data not shown) is missing 139 residues from the N-terminal, including these functionally important residues, and was rejected as a candidate to be the principal variant.

Human and mouse exonic structure for the six variants of SF1. Human exonic structure for the SF1 variants (locus AP001462.5) shown left, mouse exonic structure on the right. Differences between human and mouse exonic structure are shown as white blocks and flagged by arrows. For variants 001 and 005 (the variant that codes for the SwissProt display sequence) exonic structure is not conserved.

Sequence to structure mapping. Human neurexin 2 is 82% sequence identical to neurexin 1 over the second LNS/LG domain. From the alignments between the two neurexins, we were able to map the sequence of the four variants of the NRXN2 gene (AP001092.3) onto the neurexin 1 structure (2h0b). The SwissProt display sequence (from variant 001) has an insertion of 15 residues relative to the structure of neurexin 1. These 15 residues would need to be squeezed in between the residues marked in red and yellow on the structure, thus breaking a beta sheet. This variant is therefore unlikely to be the primary variant.

The numbers of variants discounted by each of the first four pipeline methods, showing where variants were rejected by more than one method.