Locus specific considerations

Due to the inherent differences in the processing and/or orientation of IGH, IGK, IGL and CSR, each of the analysis performed by IgCaller have some peculiarities that need to be taken into account for a proper analysis.

The IGK locus, in addition to the V and J genes, also includes the so-called intronic recombination signal sequence (RSS, downstream of the J genes) and the kappa deleting element (Kde, found 24 kb downstream of the constant K region). Once an unproductive IGK rearrangement is formed, a deletion within the Kde and RSS may occur to eliminate the constant and enhancer region of the IGK preventing the expression of the unproductive rearrangement. Similarly, a deletion involving the Kde and a given V gene may occur to completely eliminate the unproductive rearrangement²⁸. Both Kde-RSS and Kde-V gene deletions are investigated by IgCaller to further characterize this locus. Moreover, although the IGKJ genes and proximal IGKV genes are oriented on the negative strand, the IGKV4-1, IGKV5-2, and IGKV genes within the distal cluster are inverted and therefore oriented on the positive strand. Thus, rearrangements involving the latter IGKV genes occur by inversions of the IGKV genes rather than deletions.

Regarding the IGL locus it is important to notice that all V and J genes are oriented in the forward orientation relative to the genome build while IGH is oriented in the reverse orientation.

For class or isotype switching, IgCaller searches for deletions within the described CSR of IGHM and IGHA1/2, IGHE or IGHG1/2/3/4 (Huebschmann et al., in preparation)²⁶. After an inspection of the CSR regions we observed that split reads may confound a proper identification of the deletions within the CSR due to their repetitive nature, high similarities within CSR regions, and presence of a remarkable number of variants and/or sequencing artifacts. Therefore, the use of split reads in CSR analysis hindered a robust identification of the exact break points and added noise to the general identification of the potential deletions. As the exact break points within the CSR are not required to properly assess the isotype switch, we decided to exclude split reads in this specific analysis. However, in addition to determine a potential deletion by abnormal insert size reads, the coverage of a 1500 bp window upstream and 1500 bp window downstream of the CSR identified is compared to assess for a drop of coverage within the deleted region, which will emphasize the presence of the CSR. For example, if a deletion occurs within the IGHM and IGHG1, a significant reduction of coverage (or read depth) should be observed in the region within the IGHM-IGHG1. Therefore, following the previous example, the coverage of a 1500 bp window upstream of the CSR of the IGHG1 and a 1,500 bp window downstream of the CSR of IGHG1 is compared using a Wilcoxon test. Potential class switch deletions with no reduction of coverage are automatically removed. If the normal WGS is available, the coverage at each position of the window in the normal BAM file is subtracted to that of the tumor sample to consider for fluctuations in coverage that are sequence/region specific. Besides, as applied in the calculation of the score based on the number of reads, the reduction of the coverage is adjusted by the tumor purity, if available, to increase the sensitivity of IgCaller in samples with low tumor content. This comparison of coverage enhanced the specificity of the CSR detection.