GRCh38-EBV1complete: a human + complete type 1 Epstein-Barr virus STAR index for per-cell EBV detection in single-cell RNA-seq

A STAR index combining GRCh38 (hg38, UCSC chr naming, primary assembly plus unplaced/unlocalised scaffolds) with a complete type 1 Epstein-Barr virus genome representation, for detecting EBV in individual cells from single-cell RNA-seq. Three viral contigs, not one V01555.2 — EBV strain B95-8, type 1 (172,281 bp) EBV_DEL — the ~12 kb region B95-8 deletes relative to wild-type EBV (11,831 bp) V01555_TRJUNC — the terminal-repeat junction formed when the genome circularises as an episome (7,881 bp) B95-8 alone is the usual choice and it is incomplete. The deleted region carries part of the BART locus — the only polyadenylated EBV transcript in latency 0/I, the state most latently infected B cells occupy — and without the terminal-repeat contig, episome-spanning reads have nowhere to map. The host genome is not optional Against an EBV-only index, human reads are force-mapped onto the 172 kb viral genome: hypermutated IGH lands on the EBNA2 GC repeat, and EBV calls inflate in proportion to plasmablast content. This is documented, not theoretical. One analysis using an EBV-only index reported 1,596 EBV+ cells in a sample where three correct alignments find none, with a median 305 EBV UMI per "positive" cell against 3 in an actual EBV+ Burkitt lymphoma line (Raji). Known limitations Type 1 only. No type 2 (AG876) sequence. Type 2 samples will undercount EBNA2 and EBNA3 substantially. Cross-deletion junctions cannot be represented. EBV_DEL is a separate contig and a CIGAR N operation cannot span contigs, so a transcript splicing from V01555.2 across the deleted region is unrepresentable. This is structural, not a threshold. Built for 150 bp reads (--sjdbOverhang 149). Shorter reads want their own build. STAR 2.7.11b required (genome version 2.7.4a). A different STAR major version will not load this index. Contents GRCh38-EBV1complete.tar.gz — the index (~29 GB unpacked) GRCh38-EBV1complete.md5 — per-file checksums, all 16 members GRCh38-EBV1complete.tar.gz.md5 — archive checksum genomeParameters.txt — exact STAR build parameters ebv_contigs.txt — EBV contig names and lengths Verify an unpacked index against the per-file checksums. A subtly wrong reference does not raise an error — it returns plausible numbers. Usable with any STAR-based pipeline. rebveal downloads and verifies it automatically via rebveal fetch-index.