Virome studies aim to explore all viruses present in a particular environment, such as the human gut. However, one major obstacle in analyzing gut viromes, especially in infants, is contamination from external sources during sample preparation. A recent study led by researchers from the University of Groningen highlights the extent and impact of such contamination on early-life virome research.

The research team analyzed over 1,300 gut virome samples along with 55 negative controls, which are samples prepared without biological material to detect contamination. They found that more than half of the gut samples shared at least one virus strain with these controls. These overlapping sequences, referred to as the “negativeome,” suggest that contamination can occur frequently and may skew the results of virome analysis.

Comparison of sequencing depth, richness, and viral community composition
between negative controls and samples

a Number of clean reads in negative controls (NCs) vs samples. b Viral operational taxonomic unit (vOTU) richness in NCs vs samples. In (a, b), data are shown for 1313 samples and 55 NCs, distributed by study as follows: Garmaeva et al. (NCs = 1, samples = 205); Liang et al. (NCs = 38, samples = 383); Maqsood et al. (NCs = 8, samples = 78); Shah et al. (NCs = 8, samples = 647). c, d Non-metric multidimensional scaling analysis utilizing Bray–Curtis dissimilarity, computed at: c the vOTU level; d predicted host level. In (c), data are shown for 33 NCs and 1228 samples: infants <5 months (n = 372), infants >5 months (n = 710), mothers (n = 146). In (d), data are shown for 25 NCs and 1147 samples: infants <5 months (n = 290), infants >5 months (n = 711), and mothers (n = 146). For (c and d), a list of outliers not depicted in the plots can be found in Supplementary Data. In (a–d), every dot is a sample, and the dot color indicates age: infant samples (age <5 months) in yellow, infant samples (age > 5 months) in orange, maternal samples in red, and NCs in dark blue. In (a–d), boxplots visualize the median, hinges (25th and 75th percentiles), and whiskers extending up to 1.5 times the interquartile range from the hinges. Asterisks denote Benjamini–Hochberg-adjusted statistical significance values: *FDR   <   0.050, ***FDR   <   0.001, ns not significant. ‘NA’ is used when significance cannot be calculated.

Although the median level of contamination in these samples was only about 1%, some samples had contamination levels as high as 99%. The problem was more significant in infant samples than in samples from mothers. This is especially important because RNA sequencing is often used to study the virome in low-biomass samples, where even small amounts of contamination can strongly influence the data.

To address this, the researchers developed a method to assess contamination using a public database of sequences identified in negative controls. They also introduced a strain-level cleaning process that can help improve the accuracy of future virome studies.

This research underscores the importance of rigorous quality control in RNA sequencing studies of the virome and provides valuable tools for identifying and removing contamination, especially in sensitive early-life samples.