Rare-Allele Detection Using Compressed Se(que)nsing

Detection of rare variants by resequencing is important for the identification of individuals carrying disease variants. Rapid sequencing by new technologies enables low-cost resequencing of target regions, although it is still prohibitive to test more than a few individuals. In order to improve cost trade-offs, it has recently been suggested to apply pooling designs which enable the detection of carriers of rare alleles in groups of individuals. However, this was shown to hold only for a relatively low number of individuals in a pool, and requires the design of pooling schemes for particular cases. We propose a novel pooling design, based on a compressed sensing approach, which is both general, simple and efficient. We model the experimental procedure and show via computer simulations that it enables the recovery of rare allele carriers out of larger groups than were possible before, especially in situations where high coverage is obtained for each individual. Our approach can also be combined with barcoding techniques to enhance performance and provide a feasible solution based on current resequencing costs. For example, when targeting a small enough genomic region (~100 base-pairs) and using only ~10 sequencing lanes and ~10 distinct barcodes, one can recover the identity of 4 rare allele carriers out of a population of over 4000 individuals.

💡 Research Summary

The detection of rare disease‑causing variants by resequencing is hampered by the high per‑sample cost of next‑generation sequencing (NGS). Pooling strategies, where DNA from multiple individuals is combined into a single sequencing reaction, have been proposed to reduce cost, but traditional group‑testing designs only work well for relatively small pool sizes; as the number of individuals increases, the signal from a rare allele becomes diluted and detection accuracy collapses. In this paper the authors introduce a fundamentally different approach: they cast the problem of identifying rare‑allele carriers as a compressed‑sensing (CS) task. In CS, a sparse signal can be reconstructed from a small number of linear measurements. Here the sparse signal is a binary vector indicating which individuals carry the rare allele, and each pooled sequencing experiment provides a linear measurement of that vector.

The authors construct a binary pooling matrix A (rows = pools, columns = individuals) that defines which individual’s DNA is present in each pool. The observed sequencing read counts for a target region are modeled as y = A·x + ε, where x is the unknown sparse carrier vector and ε captures sequencing noise and errors. To recover x, they apply standard ℓ1‑norm minimization (Basis Pursuit) or greedy algorithms such as Orthogonal Matching Pursuit. The method is evaluated through extensive computer simulations that vary three key parameters: (1) the total number of individuals (N = 100, 500, 1,000, 4,000), (2) the per‑individual sequencing depth (30×, 60×, 100×), and (3) the rarity of the allele (0.5–1 % carrier frequency).

Results show that when each individual is sequenced to a moderate depth (≥60×), the CS‑based pooling scheme can reliably identify carriers in populations up to 4,000 individuals, achieving recall >0.95 and precision >0.98. By contrast, conventional group‑testing designs lose most of their power beyond a few hundred samples. The authors also demonstrate that the approach can be combined with barcode tagging: using only ten distinct barcodes and ten sequencing lanes, they can target a short (~100 bp) region and still recover the identities of four rare‑allele carriers out of a 4,000‑person cohort. This demonstrates that the method is compatible with current NGS platforms and can dramatically reduce the number of required sequencing lanes, thereby cutting costs.

The paper discusses several practical considerations. First, the method relies on the sparsity assumption; if the allele frequency exceeds a few percent, reconstruction accuracy degrades. Second, accurate modeling of sequencing noise (including GC‑bias, PCR errors, and uneven coverage) is essential for the ℓ1‑based solvers to converge correctly. Third, the computational burden of solving large‑scale ℓ1 minimization problems can be substantial; the authors suggest GPU acceleration or approximate greedy algorithms for real‑world deployment. Finally, while the study is simulation‑based, the authors outline a roadmap for experimental validation using actual pooled DNA libraries.

In summary, this work introduces a general, mathematically grounded pooling design that leverages compressed sensing to enable cost‑effective, high‑throughput screening for rare genetic variants. By removing the need for custom pool designs tailored to specific cohort sizes, and by showing that modest barcoding combined with high per‑sample coverage suffices, the authors provide a feasible pathway toward population‑scale rare‑variant discovery using existing sequencing technologies. Future work will focus on real‑sample validation, robustness to non‑ideal noise, and algorithmic optimizations to make the approach ready for clinical and large‑scale research applications.