A fragment library was prepared from the total DNA using the KAPA HyperPlus kit (Roche, Switzerland), according to the manufacturer's instructions. DNA was fragmented using Fragmetase in a range of lengths of 150-220 bp. After Pre-capture LM-PCR library concentration was measured by Qubit (ThermoFisher Scientific, USA) according manufacturer’s instructions. The library size and potential existence of primer/adapter dimers were determined by Agilent High Sensitivity DNA Kit (Agilent, USA), the optimal library length was 290-330 bp. Prepared libraries were mixed in 24-96 pcs in one pool, after that two rounds of the hybridization were carried out with SeqCap EZ Choice panel probes according manufacturer’s instructions. Hybridization was carried out at 47°C for 16 hours. The hybrid complexes were captured with SeqCap Capture beads, washed from non-specific fragments and amplified using KAPA HiFi HS MasterMix (Roche, Switzerland) for 5 cycles. After that, the hybridization procedure was repeated as described above. The final Post-Capture LM-PCR of enriched libraries was 16 cycles. The sequencing of the captured library pool was performed by Miseq (Illumina, USA), using 2x150 bp pair-end reads.
Sequencing data are analyzed in accordance with recommendations of the GATK Best Practices (Broad Institute) to search germline mutations. The universal pipeline scheme is shown in the figure below:
Each block of calculations is performed in an isolated environment with allocation of the optimal amount of resources and the maximum parallelization of processes. Post-processing is carried out using both individual quality filters and use of a pre-trained neural network (Convolutional Neural Network), which can significantly improve sensitivity and specificity indices for detecting mutations.