# BlCB melanoma cell line panel – DepMap‑harmonized somatic mutations (v1.0)

Release date: 2026‑05‑14  
Data version: v1.0  
Contact: e.kosobokova@ronc.ru 

## 1. Release contents

This README describes the structure of the release, file formats, and data processing workflow.  
The release includes two types of genomic data for the BlCB melanoma cell line panel:

- DepMap-harmonized somatic SNV/indel variants, prepared according to the logic of the DepMap OmicsSomaticMutations pipeline (DepMap Public 25Q3);

- a CNV table derived from WES data and annotated with OncoKB.

### 1.1 SNV/indel:

BlCB_DepMap_harmonized_public_v1.0.maf.gz — annotated MAF file containing DepMap-harmonized somatic SNV/indel variants (core and rescue) with functional, population, and clinical annotation (gnomAD, OncoKB, Cancer Hotspots, CHASMplus, CIViC, HESS).

BlCB_SNVs_DepMap_harmonized_v1.0.tsv — simplified SNV/indel table in TSV format, intended for rapid inspection and integration with other datasets.

BlCB_SNVs_DepMap_harmonized_v1.0.xlsx — Excel version of the SNV table with sheets SNVs and Data_dictionary, containing descriptions of all columns.

### 1.2 CNV:

BlCB_CNV_OncoKB_v1.0.tsv — annotated CNV table in TSV format for machine processing and downstream integration, containing segmented copy-number profiles normalized to modal ploidy.

BlCB_CNV_OncoKB_v1.0.xlsx — Excel version of the CNV table with sheets `CNV` and `Data_dictionary`, containing descriptions of all columns.

## 2. WES sequencing and processing

### 2.1. Sequencing and primary processing

- Sequencing platform: **MGISEQ‑2000** (DNBSEQ‑G400RS High‑throughput Sequencing Set PE100, MGI Tech, China).  
- Library preparation: MGIEasy Universal DNA Library Prep Set + SureSelect Human All Exon v7 (Agilent).  
- Basecalling and FASTQ generation: **zebracallV2** (MGI Tech).

WES data processing was performed according to GATK Best Practices (“Data Pre-processing for Variant Discovery”, “Somatic Short Variant Discovery (SNVs + Indels)”):

- Reference genome: **hg38/GRCh38**.  
- Alignment: **bwa‑mem2 v2.2.1**.  
- GATK: **GATK 4.5.0.0** (MarkDuplicatesSpark, BaseRecalibrator, ApplyBQSR).  

### 2.2. Somatic SNV/indel calling

- Variant calling: **Mutect2 (GATK 4.5.0.0)** in tumor-only mode using recommended resources:
  - `--germline-resource` based on gnomAD (af-only-gnomad);  
  - additional standard Mutect2 filters.  
- Mutect2 filtering:
  - GetPileupSummaries, CalculateContamination, LearnReadOrientationModel;  
  - FilterMutectCalls (default parameters).  
- Additional QC:
  - panel of normals (PoN) filtering: variants present in PoN were removed as likely artifacts;  
  - strict coverage and allele-fraction thresholds: variants retained with `DP ≥ 10` and `VAF ≥ 0.15`.  

All somatic variants for BlCB cell lines were merged into a single annotated MAF file, which was subsequently used for DepMap-harmonized filtering.

### 2.3. Annotation

Primary annotation:

- **Funcotator (GATK)** for hg38.

Additional functional and clinical annotation (via **OpenCravat** and external resources):

- gnomAD v2.1  
- Cancer Hotspots 1.2.1  
- CHASMplus SKCM 1.3.0  
- REVEL (database version as of 2024‑07‑30)  
- ClinVar (database version as of 2026‑04‑01)  
- CIViC (database version as of 2026‑02‑01)
- **OncoKB** (API v1, current version at the time of annotation)

Additional sources:

- COSMIC Cancer Gene Census (CGC, version 103, GRCh38)  
- HESS driver catalog derived from the OmicsSomaticMutations.csv file (DepMap Public 25Q3). Variants with HessDriver == TRUE were extracted and matched to our MAF by coordinates and alleles, enabling assignment of the is_hess_var and is_hess_gene flags.

## 3. DepMap-/CCLE-harmonized filtering of somatic variants

Filtering was implemented in R and, where possible, follows the logic of DepMap OmicsSomaticMutations (DepMap Public 25Q3), adapted for BlCB. The main steps used to generate the DepMap-harmonized somatic variant set are listed below.

### 3.1. Functional classes

The functional filter excludes variants with `Variant_Classification` in:

- `Silent`  
- `RNA`  
- `Intron`  
- `5'UTR`  
- `3'UTR`  
- `3'Flank`  
- `5'Flank`  
- `Splice_Region`  
- `IGR`  
- `COULD_NOT_DETERMINE`  

In addition, records with `NA` in `Hugo_Symbol` are removed.  
As a result, only potentially functional somatic classes are retained (such as `Missense_Mutation`, `Nonsense_Mutation`, frameshift/in-frame indels, `Splice_Site`, and others).

### 3.2. HESS, COSMIC, OncoKB, Hotspots, CHASMplus, CIViC annotation

At the annotated MAF level, the following flags were computed:

- `is_hess_var`, `is_hess_gene` — based on the HESS variant table as described above.  
- `is_cgc_tier1_gene` — genes from COSMIC CGC Tier 1 (v103, GRCh38).  
- `is_oncokb_oncogenic` — `ONCOGENIC ∈ {Oncogenic, Likely Oncogenic}` (OncoKB).  
- `is_cosmic_strong` — COSMIC variants with `OC_cosmic__variant_count ≥ 10` in Tier 1 CGC genes.  
- `is_cancer_hotspot` — records with non-empty `OC_cancer_hotspots__samples` (Cancer Hotspots).  
- `is_chasmplus_hit` — CHASMplus_SKCM variants with `OC_chasmplus_SKCM__pval < 0.05`.  
- `is_civic_supported` — CIViC-annotated variants (non-zero `OC_civic__molecular_profile_score` and valid `OC_civic__id`).  

A variant was considered a rescue candidate if it satisfied at least one of the following criteria:  
(i) it was located in a COSMIC CGC Tier 1 gene;  
(ii) it matched a HESS-predicted driver variant or was found in a HESS driver gene;  
(iii) it was annotated in OncoKB as oncogenic or likely oncogenic;  
(iv) it belonged to Cancer Hotspots.

### 3.3. DepMap-harmonized somatic variant set

For variants passing the functional filter (see Section 3.1), the following layers were defined:

- **Core layer**:
  - variants satisfying `gnomad_af_global ≤ 1e‑5` or `NA`.

- **Rescue layer**:
  - variants with `rescue_candidate = TRUE`;
  - satisfying `1e‑5 < gnomad_af_global < 1e‑3`.

Core and rescue variants were merged into a unified DepMap-harmonized dataset, after which duplicates were removed based on `(Chromosome, Start_Position, Reference_Allele, Tumor_Seq_Allele2, Tumor_Sample_Barcode)`.  
The resulting somatic variant set served as the basis for the annotated MAF and SNV table.

## 4. Annotated MAF and SNV table

The file `BlCB_DepMap_harmonized_public_v1.0.maf.gz` contains only variants that passed the functional filter (see Section 3.1). It includes the main coordinate fields, `VAF`/`DP`, `gnomAD`, `OncoKB`, as well as the flags `core`/`rescue`/`HESS`/`CGC`/`CancerHotspots`/`CHASMplus`/`CIViC`.

Based on the same set, the following files were generated:

- `BlCB_SNVs_DepMap_harmonized_v1.0.tsv`;
- `BlCB_SNVs_DepMap_harmonized_v1.0.xlsx`.

The SNV table contains, for each mutation:

- sample identifier (`sample_id`, `Tumor_Sample_Barcode`);
- chromosome and coordinates (`chromosome`, `start`, `end`, `ref`, `alt`, `variant_id`);
- gene symbol and functional annotation (`gene`, `consequence`, `transcript`, `protein_change`);
- `VAF` and coverage depth (`DP`);
- Mutect2 filter status (`filter_pass`, `AS_FilterStatus`);
- global gnomAD allele frequency (`population_AF`, `gnomad_af_global`);
- OncoKB annotation (`OncoKB_annotation`, `OncoKB_level`);
- integrated prioritization flag (`prioritized_flag`), based on OncoKB, CHASMplus, and REVEL.

Detailed column descriptions are provided on the `Data_dictionary` sheet of `BlCB_SNVs_DepMap_harmonized_v1.0.xlsx`.

## 5. Copy-number assessment, discretization, and CNV annotation

CNV analysis was performed using **CNVkit v0.9.13**, resulting in `seg` and `cnn` segmentation files. Copy number values for the BlCB panel were normalized to modal ploidy determined from karyotyping of the cell lines.

The CNV table contains, for each gene, the corresponding segment with copy-number change, as well as segmentation parameters:
- `log2` — log2 ratio of segment read depth relative to the diploid level;
- `cn` — absolute copy number normalized to modal ploidy;
- `depth` — raw segment depth;
- `probes` — number of probes/bins included in the segment;
- `weight` — segment weight in segmentation (CBS algorithm).

Because the OncoKB CnaAnnotator v3.4 API operates on discrete CNV states, copy-number values were mapped to the following scale:

| Condition | Discrete value | Interpretation |
|---|---:|---|
| `cn = 0` | `-2` | `deletion` |
| `log2 < -1` | `-1` | `loss` |
| `-1 ≤ log2 < 0` | `0` | `neutral` |
| `0 ≤ log2 < 1` | `1` | `gain` |
| `log2 ≥ 1` | `2` | `amplification` |

CNV interpretation was performed using **OncoKB CnaAnnotator API v3.4**. The CNV table further includes OncoKB annotation fields reflecting the functional and clinical interpretation of the discretized CNV states.

The CNV table includes the following annotation fields:
- `ALTERATION` — discretized CNV type;
- `GENE_IN_ONCOKB` — presence of the gene in OncoKB;
- `VARIANT_IN_ONCOKB` — presence of the variant in OncoKB;
- `MUTATION_EFFECT` — predicted functional effect of the variant;
- `MUTATION_EFFECT_CITATIONS` — sources used to support the effect assessment;
- `ONCOGENIC` — oncogenicity of the variant;
- `LEVEL_1`, `LEVEL_2`, `LEVEL_3A`, `LEVEL_3B`, `LEVEL_4`, `LEVEL_R1`, `LEVEL_R2` — levels of therapeutic relevance and resistance;
- `HIGHEST_LEVEL`, `HIGHEST_SENSITIVE_LEVEL`, `HIGHEST_RESISTANCE_LEVEL` — highest evidence levels;
- `TX_CITATIONS` — sources for therapeutic assertions;
- `LEVEL_Dx1`, `LEVEL_Dx2`, `LEVEL_Dx3`, `HIGHEST_DX_LEVEL`, `DX_CITATIONS` — diagnostic levels and references;
- `LEVEL_Px1`, `LEVEL_Px2`, `LEVEL_Px3`, `HIGHEST_PX_LEVEL`, `PX_CITATIONS` — prognostic levels and references.