Fungal Signature of Moisture Damage in Buildings: Identification by Targeted and Untargeted Approaches with Mycobiome Data

Living or working in damp or moldy buildings increases the risk of many adverse health effects, including asthma and other respiratory diseases. To date, however, the particular environmental exposure(s) from water-damaged buildings that causes the health effects have not been identified. Likewise, a consistent quantitative measurement that would indicate whether a building is water damaged or poses a health risk to occupants has not been found. In this work, we tried to develop analytical tools that would find a microbial signal of moisture damage amid the noisy background of microorganisms in buildings. The most successful approach taken here focused on particular groups of fungi—those considered likely to grow in damp indoor environments—and their associations with observed moisture damage. With further replication and refinement, this hypothesis-based strategy may be effective in finding still-elusive relationships between building damage and microbiomes.

Identifying a fungal signature of moisture damage in buildings by taking a targeted approach with microbiome data Rachel I. Adams, Iman Sylvain, Michal P. Spilak, John W. Taylor, Michael S. Waring, and Mark J. Mendell Figure S1 Distribution of count data Sum of hydrophilic, mesophilic, xerophilic, and ERMI Group 1 fungi in study samples, ranked from highest to lowest sum.

Hydrophilic fungi
Rank Sample sums

Mesophilic fungi
Rank Sample sums    1 Exponential of the negative binomial model estimate, interpreted as the estimated relative change in the relative abundance 2 For prediction variables with two categories (e.g. present vs. absent), the p-values were determined using the z test statistic on the model coefficient. For prediction variables with three categories (e.g., none, low, high), pvalues were determined using a two degree-of-freedom chi-square test on the full and reduced models. 3 High damage category has fewer than 5 individuals in that group 4 Bold, p-value ≤ 0.05 Table S4 Absolute abundance of fungal groups The absolute abundance of hydrophilic, mesophilic, and xerophilic fungi in vacuum and dustfall collectors for homes with and without building damage; means of the absolute abundance in each group, intepreted as gene copy equivalements, and estimated relative change from a negative binomial model, with p-values of the model coefficients.   Table 8: Relative abundance of ERMI Group 1 fungi in door trim in rooms homes with and without building damage in comparison to homes with and without building damage; means of the relative abundance in each group and estimated relative change from a negative binomial model, with p-values of the model coefficients. Comparison of the sampling that occurred contemporaneously ("winter") and following the building damage assessment ("summer") as well as with taxonomy assignment using the UNITE fungal reference database with global and reference singletons. 1 Exponential of the negative binomial model estimate, interpreted as the estimated relative change in the relative abundance 2 For prediction variables with two categories (e.g. present vs. absent), the p-values were determined using the z test statistic on the model coefficient. For prediction variables with three categories (e.g., none, low, high), pvalues were determined using a two degree-of-freedom chi-square test on the full and reduced models. 3 High damage category has fewer than 5 individuals in that group 4 Bold, p-value ≤ 0.05 Table S6 ERMI Group 1 fungi in vacuum dust and dustfall collectors Relative abundance of ERMI Group 1 fungi in vacuum samples and dustfall collectors in homes with and without building damage; means of the relative abundance in each group and estimated relative change from a negative binomial model, with p-values of the model coefficients. Comparison of the sampling that occurred contemporaneously ("winter") and following the building damage assessment ("summer") as well as with taxonomy assignment using the UNITE fungal reference database with global and reference singletons.

Comparison of UNITE global singletons versus UNITE reference singletons
In approach 1, we observed 57 species with the UNITE reference singletons fungal database. Counterintuitively, the proportion of sequences representing taxa with known aw requirements for growth was greater when taxonomy was assigned with reference singletons (35%) than when assigned with global singletons (28%) even though the UNITE database with reference singletons contains fewer sequences (~18K taxa) than the database with global singletons (~35K taxa). We attribute this to the fact that there were more taxa unresolved at the species level (i.e. "NA") when taxonomy was assigned with UNITE global singletons (n=96 taxa) than when assigned with UNITE reference singletons (n=26 taxa).
The taxonomic assignment of the sequences could differ between UNITE databases: 52 species were identified in both UNITE databases, while an additional eight species were found only when taxonomy was assigned with global singletons, and an additional five species only when taxonomy was assigned with reference singletons. Nevertheless, over 65% of the 627 ASVs were identified to the same species between the two databases and another 14% of these 627 ASVs were identified to the same genus but differed at the species-level identification between the two databases. Only 5 taxa (<1% of the ASVs) were identified to different genera by the two fungal databases. The most common example of different species assignments was that many ASVs were identified as Aspergillus versicolor using the UNITE fungal database with global singletons but as A. sydowii in the database with reference singletons. However, both of these species are considered xerophilic fungi and thus would be included in the targeted xerophilic group. Finally, 20% of the taxa were identified to a species of interest with one database but were unresolved (i.e. "NA") at the genus or species level with the other database.
In approach 2, 28 were identified in the mycobiome data with reference singletons, representing over 10% of the sequences. As with Approach 1, assigning taxonomy with reference singletons captured slightly more named species in the current dataset than assigning taxonomy with global singletons. Most Group 1 fungi were found in both fungal databases, although Aspergillus fischeri and A. foetidus were identified only in the database with global singletons, and A. sclerotiorum, Chaetomium globosum, Penicillium spinulosum, and Trichoderma koningii were identified only in the database with reference singletons.