The risks associated with re-identification of human genetic data are severely limiting open data sharing in life sciences, even in studies where donor-related genetic variant information is not of primary interest. Here, Karolinska Institute researchers developed BAMboozle, a versatile tool to eliminate critical types of sensitive genetic information in human sequence data by reverting aligned reads to the genome reference sequence. Applying BAMboozle to functional genomics data, such as single-cell RNA-seq (scRNA-seq) and scATAC-seq datasets, confirmed the removal of donor-related single nucleotide polymorphisms (SNPs) and indels in a manner that did not disclose the altered positions. Importantly, BAMboozle only removes the genetic sequence variants of the sample (i.e., donor) while preserving other important aspects of the raw sequence data. For example, BAMboozled scRNA-seq data contained accurate cell-type associated gene expression signatures, splice kinetic information, and can be used for methods benchmarking. Altogether, BAMboozle efficiently removes genetic variation in aligned sequence data, which represents a step forward towards open data sharing in many areas of genomics where the genetic variant information is not of primary interest.
Schematic overview of the BAMboozle procedure
a Sequenced reads typically harbor identifying genetic variants that are deviating from the reference genome. Data sanitation removes such variants by replacement with the reference sequence. b Schematic representation of an aligned 100 bp read (red) with an alternative allele at a single-nucleotide polymorphism (SNP) position is corrected to the reference base in output read (blue). Raw and corrected CIGAR (Compact Idiosyncratic Gapped Alignment Report) strings (with operator descriptions below) are shown over reads, and raw and corrected MD tags are shown to the right. c Aligned 100 bp read (red) with donor-specific insertion (relative to reference), with information as in b. d Aligned 100 bp read (red) with donor-specific deletion (relative to reference), with information as in b. e Aligned 100 bp read pair sequence (paired-end data) that was clipped at the 5’ end relative to reference. Note the correction in alignment end position as a result of the removal of clipped positions. f Spliced alignment that additionally contains an insertion not present in reference. The spliced alignment information is preserved during the sanitation of genetic variation. g Spliced alignment that additionally contains a deletion not present in reference. The spliced alignment information is preserved during the sanitation of genetic variation. CIGAR string operators: M, alignment match; I, insertion; D, deletion; N, skipped in reference; S, soft clipping; H, hard clipping; P, padding; =, sequence match; X, sequence mismatch.