When Less is More: “Slicing” Sequencing Data Improves Read Decoding Accuracy andDe NovoAssembly Quality

Abstract

Since the invention of DNA sequencing in the seventies, computational biologists have had to deal with the problem de novo genome assembly with limited (or insufficient) depth of sequencing. In this work, for the first time we investigate the opposite problem, that is, the challenge of dealing with excessive depth of sequencing. Specifically, we explore the effect of ultra-deep sequencing data in two domains: (i) the problem of decoding reads to BAC clones (in the context of the combinatorial pooling design proposed in [1]), and (ii) the problem of de novo assembly of BAC clones. Using real ultra-deep sequencing data, we show that when the depth of sequencing increases over a certain threshold, sequencing errors make these two problems harder and harder (instead of easier, as one would expect with error-free data), and as a consequence the quality of the solution degrades with more and more data. For the first problem, we propose an effective solution based on “divide and conquer”: we “slice” a large dataset into smaller samples of optimal size, decode each slice independently, then merge the results. Experimental results on over 15,000 barley BACs and over 4,000 cowpea BACs demonstrate a significant improvement in the quality of the decoding and the final assembly. For the second problem, we show for the first time that modern de novo assemblers cannot take advantage of ultra-deep sequencing data.

Scrible: Ultra-Accurate Error-Correction of Pooled Sequenced Reads

“…Once the set of MTP BACs has been pooled, we sequenced the resulting pools and used the read decoding algorithm HashFilter [8,9] to assign the reads back to their source BACs, and finally assemble each BAC individually. Error correction is applied prior to read decoding.…”

“…In [8] we solved the read decoding problem with a combinatorial algorithm, while in [1] we proposed a compressed sensing approach. In [9] we further improved the decoding by a "data slicing" approach. A similar slicing strategy was used in [10] to improve the assembly quality for ultra-deep sequencing data.…”

“…As the read decoding problem is presented elsewhere [1,8,9], here we only provide a brief overview to motivate the necessity of correcting reads before decoding them. Given a set of pools P and a set of BACs B, the signature for a BAC b ∈ B is the subset A ⊂ P of pools (|A| = L) to which BAC b is assigned.…”

“…After the selection of the BACs to be sequenced (see [8,9] for more details), we pool them according to a scheme that allows us to decode (assign) sequenced reads back to their corresponding BACs.…”

“…We have recently demonstrated how to take advantage of combinatorial pooling (also known as group testing) for clone-by-clone de-novo genome sequencing [1,8,9]. In our sequencing protocol, subsets of non-redundant genome-tiling BACs are chosen to form intersecting pools, then groups of pools are sequenced on an Illumina sequencing instrument via standard multiplexing (DNA barcoding).…”