Journal: Nature biotechnology

Article Title: Establishing community reference samples, data and call sets for benchmarking cancer mutation detection using whole-genome sequencing

doi: 10.1038/s41587-021-00993-6

Figure Lengend Snippet: Data from multiple NGS platforms were obtained to detect and confirm germline and somatic variants in HCC1395 and HCC1395BL cell lines

Article Snippet: HCC1395BL human B lymphoblast Epstein–Barr virus-transformed cells (expanded from ATCC CRL-2325) were cultured in ATCC-formulated Iscove’s Modified Dulbecco’s medium (ATCC, 30–2005) supplemented with FBS to a final concentration of 20%.

Techniques: Sequencing, Biomarker Discovery, Microarray

Journal: Nature biotechnology

Article Title: Establishing community reference samples, data and call sets for benchmarking cancer mutation detection using whole-genome sequencing

doi: 10.1038/s41587-021-00993-6

Figure Lengend Snippet: Twenty-one sequencing replicates for the tumor (HCC1395) and normal (HCC1395BL) gDNA samples were performed at six sequencing centers. They were grouped into five groups shown as colored squares on the far left. The sequencing platforms used were HiSeq and NovaSeq, and the sequencing experiments were performed at Illumina (IL), Fudan University (FD), Novartis (NV), European Infrastructure for Translational Medicine (EA), National Cancer Institute (NC) and Loma Linda University (LL). Each of the 21 sequencing datasets was aligned with three aligners to create a total of 63 pairs of tumor–normal BAM files. For each of the 63 tumor–normal BAM files, we ran six somatic mutation callers (MuTect2, SomaticSniper, VarDict, MuSE, Strelka2 and TNscope) followed by SomaticSeq and NeuSomatic machine learning classifiers using models built specifically for these datasets. Initial tier was assigned to each variant call based on how consistently the variant was classified as a somatic mutation across different sequencing centers and aligners. Then, to rescue low-VAF variants into the reference call set, we ran the same variant-calling pipeline on two higher-depth datasets: an independent 350× replicate sequenced on a HiSeq at Genentech (GT) and a 400× replicate by combining nine NovaSeq replicates at Illumina (IL). High-confidence calls from those two datasets were used to promote less-reproducible low-VAF variants from the 21 sequencing replicates into the reference call set. Then, we combined all of our short-read sequencing data into a pair of 1,500× tumor–normal and used it to rescue additional low-VAF variants into the reference call set. Finally, we cross-referenced our Illumina short-read-based high-confidence calls with PacBio long-read sequencing data and removed a small number of calls that were inconsistent with each other. The high-confidence calls (labeled PASS) were considered ‘true’ somatic mutations, and genomic regions with low-confidence calls were removed from the high-confidence regions. Chr, chromosome; Pos, position.

Article Snippet: HCC1395BL human B lymphoblast Epstein–Barr virus-transformed cells (expanded from ATCC CRL-2325) were cultured in ATCC-formulated Iscove’s Modified Dulbecco’s medium (ATCC, 30–2005) supplemented with FBS to a final concentration of 20%.

Techniques: Sequencing, Mutagenesis, Variant Assay, Labeling

Journal: Nature biotechnology

Article Title: Establishing community reference samples, data and call sets for benchmarking cancer mutation detection using whole-genome sequencing

doi: 10.1038/s41587-021-00993-6

Figure Lengend Snippet: a, The inferred tumor phylogenetic tree. The subclone S1 represents the most recent common ancestor (MRCA) of all tumor cells, and S2 to S10 represent the subclones with various cancer cell fractions (for example, S2: 60%). The edges represent the evolutionary relationships between subclones. Subclones S3 and S6 are not shown given that their cancer cell fractions were less than 10%. Most point mutations are in driver genes (labeled beside the edges) present in the MRCA. Using the 10x Genomics Single Cell CNV Solution, integer-scaled CNA profiles were obtained across the genomes of 638 HCC1395BL cells (b) and 1,270 HCC1395 cells (c). Noisy cells and cells in the S phase of the cell cycle were removed. The complete linkage method was used for hierarchical clustering. Each row represents a cell being sequenced; similar cells were clustered together based on CNVs. Chromosome-scale gains are in orange and losses are in blue in the heat maps.

Article Snippet: HCC1395BL human B lymphoblast Epstein–Barr virus-transformed cells (expanded from ATCC CRL-2325) were cultured in ATCC-formulated Iscove’s Modified Dulbecco’s medium (ATCC, 30–2005) supplemented with FBS to a final concentration of 20%.

Techniques: Labeling

Journal: Nature biotechnology

Article Title: Establishing community reference samples, data and call sets for benchmarking cancer mutation detection using whole-genome sequencing

doi: 10.1038/s41587-021-00993-6

Figure Lengend Snippet: (a) Karyotype of HCC1395. Cytogenetic analysis was performed on ten G-Banded metaphase cells from HCC1395. Analysis pointed to a hypertetraploid line with chromosome counts ranging from 64–79 and gain of 38–63 unidentifiable marker chromosomes. (b) Karyotype of HCC1395BL. Cytogenetic analysis was performed on ten G-banded metaphase cells from HCC1395BL. All ten cells showed loss of a chrX and an unbalanced whole arm translocation between the long-arm of chr6 at band q10 and the short-arm of chr16 at band p10. This resulted in a net loss of one copy of the short-arm of chr6 and loss of one copy of the long-arm of chr16. The abnormal chromosome could be placed in either a chr6 or chr16 locus as we were unable to determine if the centromere belongs to chr6 or chr16 (inset figure).

Article Snippet: HCC1395BL human B lymphoblast Epstein–Barr virus-transformed cells (expanded from ATCC CRL-2325) were cultured in ATCC-formulated Iscove’s Modified Dulbecco’s medium (ATCC, 30–2005) supplemented with FBS to a final concentration of 20%.

Techniques: Marker, Translocation Assay

Journal: Nature biotechnology

Article Title: Establishing community reference samples, data and call sets for benchmarking cancer mutation detection using whole-genome sequencing

doi: 10.1038/s41587-021-00993-6

Figure Lengend Snippet: Cytogenetic analysis with Affymetrix Cytoscan HD microarray. (a) Cytogenetic view of HCC1395. (b) Cytogenetic view of HCC1395BL. The losses of chr6p, chr16q, and chrX were confirmed.

Article Snippet: HCC1395BL human B lymphoblast Epstein–Barr virus-transformed cells (expanded from ATCC CRL-2325) were cultured in ATCC-formulated Iscove’s Modified Dulbecco’s medium (ATCC, 30–2005) supplemented with FBS to a final concentration of 20%.

Techniques: Microarray

Journal: Nature biotechnology

Article Title: Establishing community reference samples, data and call sets for benchmarking cancer mutation detection using whole-genome sequencing

doi: 10.1038/s41587-021-00993-6

Figure Lengend Snippet: (a) VAF of truth set germline SNVs in HCC1395BL. The copy numbers of HCC1395BL were predicted by Affymetrix Cytoscan HD microarray. (b) VAF of the truth set germline SNV positions (discovered in HCC1395BL) in HCC1395. (c) VAF of the truth set somatic SNVs in HCC1395. The copy numbers of HCC1395 were predicted by ascatNgs.

Article Snippet: HCC1395BL human B lymphoblast Epstein–Barr virus-transformed cells (expanded from ATCC CRL-2325) were cultured in ATCC-formulated Iscove’s Modified Dulbecco’s medium (ATCC, 30–2005) supplemented with FBS to a final concentration of 20%.

Techniques: Variant Assay, Microarray