Mind the gaps: overlooking inaccessible regions confounds statistical testing in genome analysis

Abstract

BackgroundThe current versions of reference genome assemblies still contain gaps represented by stretches of Ns. Since high throughput sequencing reads cannot be mapped to those gap regions, the regions are depleted of experimental data. Moreover, several technology platforms assay a targeted portion of the genomic sequence, meaning that regions from the unassayed portion of the genomic sequence cannot be detected in those experiments. We here refer to all such regions as inaccessible regions, and hypothesize that ignoring these regions in the null model may increase false findings in statistical testing of colocalization of genomic features.ResultsOur explorative analyses confirm that the genomic regions in public genomic tracks intersect very little with assembly gaps of human reference genomes (hg19 and hg38). The little intersection was observed only at the beginning and end portions of the gap regions. Further, we simulated a set of synthetic tracks by matching the properties of real genomic tracks in a way that nullified any true association between them. This allowed us to test our hypothesis that not avoiding inaccessible regions (as represented by assembly gaps) in the null model would result in spurious inflation of statistical significance. We contrasted the distributions of test statistics and p-values of Monte Carlo-based permutation tests that either avoided or did not avoid assembly gaps in the null model when testing colocalization between a pair of tracks. We observed that the statistical tests that did not account for assembly gaps in the null model resulted in a distribution of the test statistic that is shifted to the right and a distribution of p-values that is shifted to the left (indicating inflated significance). We observed a similar level of inflated significance in hg19 and hg38, despite assembly gaps covering a smaller proportion of the latter reference genome.ConclusionWe provide empirical evidence demonstrating that inaccessible regions, even when covering only a few percentages of the genome, can lead to a substantial amount of false findings if not accounted for in statistical colocalization analysis.Electronic supplementary materialThe online version of this article (10.1186/s12859-018-2438-1) contains supplementary material, which is available to authorized users.

Identifying the causes and consequences of assembly gaps using a multiplatform genome assembly of a bird‐of‐paradise

“…Scaffold N50 indicates the minimum scaffold size among the largest scaffolds making up half of the assembly, while BUSCO values measure the number of complete/incomplete/missing core genes in the assembly. However, genome completeness goes beyond scaffold N50 and gene presence (Domanska et al., 2018; Sedlazeck et al., 2018; Thomma et al., 2016). Genes usually occupy a small fraction of genomes and new sequencing technologies commonly yield high N50 values.…”

“…and gene presence (Domanska et al, 2018;Sedlazeck et al, 2018;Thomma et al, 2016). Genes usually occupy a small fraction of genomes and new sequencing technologies commonly yield high N50…”

“…It is essential to be aware of technological biases and genome assembly incompleteness during project design. A growing body of evidence suggests that these biases and assembly gaps can affect downstream analyses and mislead biological interpretations (Domanska et al., 2018; Peona et al., 2018; Thomma et al., 2016; Weissensteiner et al., 2017). For example, GC‐skewed coverage is particularly important in birds, where ~15% of genes are so GC‐rich that they are often not represented in Illumina‐based genome assemblies (Botero‐Castro et al., 2017; Hron et al., 2015).…”

The changing face of genome assemblies: Guidance on achieving high‐quality reference genomes

“…As noted above, well known deficiencies in genome assemblies include difficulty in assembling repetitive, duplicated, and GC rich regions that can often be addressed with long‐read sequencing (Sedlazeck et al, 2018), but at the expense of sequencing error which may influence estimates of gene content (Jaworski et al, 2020; Watson & Warr, 2019). The trade‐offs of various sequencing technologies can be offset by applying multiple platforms (Peona et al, 2021; Rhie, McCarthy, et al, 2020), however many chromosome‐level genome assemblies have extensive gaps within and among scaffolds that may hinder utility of these hybrid assemblies for subsequent studies (Domanska et al, 2018; Peona et al, 2021). Thus, researchers should strive to accurately report the quality of genome assemblies with key statistics that account for fragmentation, accuracy, and gene content.…”

An improved de novo genome assembly of the common marmoset genome yields improved contiguity and increased mapping rates of sequence data

“…The 2.26 Gb assembled genome from a female marmoset, although was sorted out into chromosomes, contained many shorter contigs and also 187,214 gap regions. These hard to assemble gap regions cannot be ignored, as they can lead to false positive results [8], and the gap regions could harbor many functionally relevant genes [9]. Recent studies have uncovered that many genes were wrongly labelled as missing in bird genomes, because of the locality of those genes being GCrich and hence had posed challenges in identifying them [9].…”