Figure 3: Germline filter.

(a) Schematic of types of sequences present following clustering and consensus building. True germline sequences are found in antibody sequences that share an identical V region and are present in multiple unique rearrangements as shown by a high number of unique CDR3 segments. Non-germline sequences include those that have undergone SHM at activation-induced cytosine deaminase (AID) hotspots and, finally, those that are overrepresented owing to PCR bias issues. The application of the germline filter leaves only germline sequences remaining in the final output. (b) Illustration of clustal analysis of a section of candidate germline VH alleles from the H1 human IgM library before germline filtering. Each sequence was compared with the full IMGT human VH database and those identical to germline genes are indicated with a red arrow. The relative frequency of V usage, based on unique CDR3 content, is indicated by the number following the ‘R’ of each sequence cluster (R1 being the highest frequency). (c) Results of the application of the pregermline and germline filters following a single iteration analysis of the H1 human IgM library. Candidate VH sequences (n=1152), were produced via consensus building using clusters identified by Windowed and Linkage cluster analysis (see Methods). To illustrate the effect of the germline filter sequences were screened against the current (August 2016) IMGT human VH database to identify the proportion showing 100% identity to validated IMGT germline sequences. This analysis was repeated with the resulting outputs of the pregermline and finally the germline filter.