Figure 2

Processes whereby reads from one gene can lead to complex contigs. a) Alignment with pre-mRNA sequences. A gene sequence, including five and three prime untranslated regions, exons and introns, is initially transcribed completely from DNA into pre messenger RNA (pre-mRNA). Amplification and sequencing of RNA libraries may therefore generate contigs which represent pre-mRNA variants (some or all introns are included). In this example, variant 1 contains sequence from the three exons and both introns (solid lines) while variant 2 contains sequence from the three exons but only the first intron (solid line, the second intron is absent, represented by a thin dashed line). The same gene sequence may therefore be assembled into more than one contig. Alignment of such contigs is likely to occur across the whole length of one of the contigs, but occur disjointly for the other (in this case, the second intron in variant 1 cannot align to variant 2, so the two alignments between variant 1 and variant 2 occurs across the whole length of variant 2 but not variant 1) b) Alignment and detection of splice variants. Reads generated from two different splice variants are shown in the middle of the figure; reads "missing" an exon are discontinuous across the whole coding sequence. Bottom left: all reads are (incorrectly) assembled into a single contig, with reads from the second splice variant contributing to 'strings' of polymorphism (false SNPs) in the consensus sequence (regions shaded grey). Bottom right: reads from the two splice variants are assembled into separate contigs, with no regions of poor sequence similarity