Kristoffer Sahlin

The Norway spruce genome sequence and conifer genome evolution

Conifers have dominated forests for more than 200 million years and are of huge ecological and economic importance. Here we present the draft assembly of the 20-gigabase genome of Norway spruce (Picea abies), the first available for any gymnosperm. The number of well-supported genes (28,354) is similar to the >100 times smaller genome of Arabidopsis thaliana, and there is no evidence of a recent whole-genome duplication in the gymnosperm lineage. Instead, the large genome size seems to result from the slow and steady accumulation of a diverse set of long-terminal repeat transposable elements, possibly owing to the lack of an efficient elimination mechanism. Comparative sequencing of Pinus sylvestris, Abies sibirica, Juniperus communis, Taxus baccata and Gnetum gnemon reveals that the transposable element diversity is shared among extant conifers. Expression of 24-nucleotide small RNAs, previously implicated in transposable element silencing, is tissue-specific and much lower than in other plants. We further identify numerous long (>10,000 base pairs) introns, gene-like fragments, uncharacterized long non-coding RNAs and short RNAs. This opens up new genomic avenues for conifer forestry and breeding.

BESST - Efficient scaffolding of large fragmented assemblies

BackgroundThe use of short reads from High Throughput Sequencing (HTS) techniques is now commonplace in de novo assembly. Yet, obtaining contiguous assemblies from short reads is challenging, thus making scaffolding an important step in the assembly pipeline. Different algorithms have been proposed but many of them use the number of read pairs supporting a linking of two contigs as an indicator of reliability. This reasoning is intuitive, but fails to account for variation in link count due to contig features.We have also noted that published scaffolders are only evaluated on small datasets using output from only one assembler. Two issues arise from this. Firstly, some of the available tools are not well suited for complex genomes. Secondly, these evaluations provide little support for inferring a software’s general performance.ResultsWe propose a new algorithm, implemented in a tool called BESST, which can scaffold genomes of all sizes and complexities and was used to scaffold the genome of P. abies (20 Gbp). We performed a comprehensive comparison of BESST against the most popular stand-alone scaffolders on a large variety of datasets. Our results confirm that some of the popular scaffolders are not practical to run on complex datasets. Furthermore, no single stand-alone scaffolder outperforms the others on all datasets. However, BESST fares favorably to the other tested scaffolders on GAGE datasets and, moreover, outperforms the other methods when library insert size distribution is wide.ConclusionWe conclude from our results that information sources other than the quantity of links, as is commonly used, can provide useful information about genome structure when scaffolding.

Improved gap size estimation for scaffolding algorithms

MOTIVATION One of the important steps of genome assembly is scaffolding, in which contigs are linked using information from read-pairs. Scaffolding provides estimates about the order, relative orientation and distance between contigs. We have found that contig distance estimates are generally strongly biased and based on false assumptions. Since erroneous distance estimates can mislead in subsequent analysis, it is important to provide unbiased estimation of contig distance. RESULTS In this article, we show that state-of-the-art programs for scaffolding are using an incorrect model of gap size estimation. We discuss why current maximum likelihood estimators are biased and describe what different cases of bias we are facing. Furthermore, we provide a model for the distribution of reads that span a gap and derive the maximum likelihood equation for the gap length. We motivate why this estimate is sound and show empirically that it outperforms gap estimators in popular scaffolding programs. Our results have consequences both for scaffolding software, structural variation detection and for library insert-size estimation as is commonly performed by read aligners. AVAILABILITY A reference implementation is provided at https://github.com/SciLifeLab/gapest. SUPPLEMENTARY INFORMATION Supplementary data are availible at Bioinformatics online.

Assembly scaffolding with PE-contaminated mate-pair libraries

MOTIVATION Scaffolding is often an essential step in a genome assembly process, in which contigs are ordered and oriented using read pairs from a combination of paired-end libraries and longer-range mate-pair libraries. Although a simple idea, scaffolding is unfortunately hard to get right in practice. One source of problems is so-called PE-contamination in mate-pair libraries, in which a non-negligible fraction of the read pairs get the wrong orientation and a much smaller insert size than what is expected. This contamination has been discussed before, in relation to integrated scaffolders, but solutions rely on the orientation being observable, e.g. by finding the junction adapter sequence in the reads. This is not always possible, making orientation and insert size of a read pair stochastic. To our knowledge, there is neither previous work on modeling PE-contamination, nor a study on the effect PE-contamination has on scaffolding quality. RESULTS We have addressed PE-contamination in an update to our scaffolder BESST. We formulate the problem as an integer linear program which is solved using an efficient heuristic. The new method shows significant improvement over both integrated and stand-alone scaffolders in our experiments. The impact of modeling PE-contamination is quantified by comparing with the previous BESST model. We also show how other scaffolders are vulnerable to PE-contaminated libraries, resulting in an increased number of misassemblies, more conservative scaffolding and inflated assembly sizes. AVAILABILITY AND IMPLEMENTATION The model is implemented in BESST. Source code and usage instructions are found at https://github.com/ksahlin/BESST BESST can also be downloaded using PyPI. CONTACT ksahlin@kth.se SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

Gap Filling as Exact Path Length Problem

One of the last steps in a genome assembly project is filling the gaps between consecutive contigs in the scaffolds. This problem can be naturally stated as finding an \(s\)-\(t\) path in a directed graph whose sum of arc costs belongs to a given range (the estimate on the gap length). Here \(s\) and \(t\) are any two contigs flanking a gap. This problem is known to be NP-hard in general. Here we derive a simpler dynamic programming solution than already known, pseudo-polynomial in the maximum value of the input range. We implemented various practical optimizations to it, and compared our exact gap filling solution experimentally to popular gap filling tools. Summing over all the bacterial assemblies considered in our experiments, we can in total fill 28% more gaps than the best previous tool and the gaps filled by our method span 80% more sequence. Furthermore, the error level of the newly introduced sequence is comparable to that of the previous tools.

Correcting bias from stochastic insert size in read pair data—applications to structural variation detection and genome assembly

Insert size distributions from paired read protocols are used for inference in bioinformatic applications such as genome assembly and structural variation detection. However, many of the models that are being used are subject to bias. This bias arises when we assume that all insert sizes within a distribution are equally likely to be observed, when in fact, size matters. These systematic errors exist in popular software even when the assumptions made about data are true. We have previously shown that bias occurs for scaffolders in genome assembly. Here, we generalize the theory and demonstrate that it is applicable in other contexts. We provide examples of bias in state-of the-art software and improve them using our model. One key application of our theory is structural variation detection using read pairs. We show that an incorrect null-hypothesis is commonly used in popular tools and can be corrected using our theory. Furthermore, we approximate the smallest size of indels that are possible to discover given an insert size distribution. Two other applications are inference of insert size distribution on de novo genome assemblies and error correction of genome assemblies using mated reads. Our theory is implemented in a tool called GetDistr (https://github.com/ksahlin/GetDistr).

Genome scaffolding with PE-contaminated mate-pair libraries

Scaffolding is often an essential step in a genome assembly process, in which contigs are ordered and oriented using read pairs from a combination of paired-ends libraries and longer-range mate-pair libraries. Although a simple idea, scaffolding is unfortunately hard to get right in practice. One source of problem is so-called PE-contamination in mate-pair libraries, in which a non-negligible fraction of the read pairs get the wrong orientation and a much smaller insert size than what is expected. This contamination has been discussed in previous work on integrated scaffolders in end-to-end assemblers such as Allpaths-LG and MaSuRCA but the methods relies on the fact that the orientation is observable, e.g., by finding the junction adapter sequence in the reads. This is not always the case, making orientation and insert size of a read pair stochastic. Furthermore, work on modeling PE-contamination has so far been disregarded in stand-alone scaffolders and the effect that PE-contamination has on scaffolding quality has not been examined before. We have addressed PE-contamination in an update of our scaffolder BESST. We formulate the problem as an Integer Linear Program (ILP) and use characteristics of the problem, such as contig lengths and insert size, to efficiently solve the ILP using a linear amount (with respect to the number of contigs) of Linear Programs. Our results show significant improvement over both integrated and standalone scaffolders. The impact of modeling PE-contamination is quantified by comparison with the previous BESST model. We also show how other scaffolders are vulnerable to PE-contaminated libraries, resulting in increased number of misassemblies, more conservative scaffolding, and inflated assembly sizes. The model is implemented in BESST. Source code and usage instructions are found at https://github.com/ksahlin/BESST. BESST can also be downloaded using PyPI.

Structural Variation Detection with Read Pair Information: An Improved Null Hypothesis Reduces Bias

Reads from paired-end and mate-pair libraries are often utilized to find structural variation in genomes, and one common approach is to use their fragment length for detection. After aligning read pairs to the reference, read pair distances are analyzed for statistically significant deviations. However, previously proposed methods are based on a simplified model of observed fragment lengths that does not agree with data. We show how this model limits statistical analysis of identifying variants and propose a new model by adapting a model we have previously introduced for contig scaffolding, which agrees with data. From this model, we derive an improved null hypothesis that when applied in the variant caller CLEVER, reduces the number of false positives and corrects a bias that contributes to more deletion calls than insertion calls. We advise developers of variant callers with statistical fragment length-based methods to adapt the concepts in our proposed model and null hypothesis.

Structural Variation Detection with Read Pair Information—An Improved Null-Hypothesis Reduces Bias

Reads from paired-end and mate-pair libraries are often utilized to find structural variation in genomes, and one common approach is to use their fragment length for detection. After aligning read-pairs to the reference, read-pair distances are analyzed for statistically significant deviations. However, previously proposed methods are based on a simplified model of observed fragment lengths that does not agree with data. We show how this model limits statistical analysis of identifying variants and propose a new model, by adapting a model we have previously introduced for contig scaffolding, which agrees with data. From this model we derive an improved null hypothesis that, when applied in the variant caller CLEVER, reduces the number of false positives and corrects a bias that contributes to more deletion calls than insertion calls. A reference implementation is freely available at https://github.com/ksahlin/GetDistr.