Abstract
Environmental exposures can shape microbial community compositions inside homes. Metagenomic sequencing methods can further elucidate the role of household exposures like indoor moisture and the surrounding landscape. To identify household environmental exposures associated with the house dust metagenome. Microbial communities in vacuumed dust from 771 homes in the Agricultural Lung Health Study were characterized using whole metagenome shotgun sequencing (5,821 taxa across 45 phyla). Household characteristics (i.e. presence of leaks, de-humidifier, humidifier use) were assessed by questionnaires or field technicians. We evaluated associations between exposures and both overall microbial diversity and differentially abundant taxa (ANCOM-BC2). Additionally, we explored microbial networks based on Spearman correlations (SECOM). Microbial diversity was higher in homes with mold/mildew (p-value<0.05), leaks, humidifier use, or occupants removing shoes before entering (p-value<0.1). Examining individual species, <10 taxa were significantly differentially abundant (p-value<0.05 after Holm-Bonferroni correction) in relation to both mold/mildew and leaks. Greater than 10 species were significantly differentially abundant in relation to removing shoes and humidifier use. Additionally, the genera Clostridium, Prevotella, and Cryptobacteroides were positively associated with removing shoes. In this farming population, the house dust microbiome differed by moisture-related exposures, and removing shoes before entering the home. Many novel associations were identified between individual taxa and these exposures. Our findings further knowledge of the impact of environmental conditions inside the home on the indoor microbiome.
Keywords: built environment, whole genome sequencing, indoor microbiome, microbial ecology, house dust microbiome
Graphical Abstract
1. Introduction
The house dust microbiome is composed of a diverse array of microorganisms. Exposure to microbial communities is known to influence human health, both positively and negatively. Childhood exposure to greater microbial diversity may lower rates of allergic and inflammatory disease [1, 2]. Specific taxa have been associated, in both positive and negative directions, with allergies, atopy, asthma, and lung health [1, 3]. Given the evolving discovery of health impacts, it is of interest to understand what factors influence the house dust microbiome.
The outdoor and built environment within the home is known to influence the house dust microbiome through a variety of exposures. Home occupancy, including humans, pets, and insects, can alter the microbiome composition and increase diversity [4, 5]. Additionally, lifestyle choices and behaviors can influence the microbiome. For example, cleaning was found to be associated with lower microbiome diversity and altered household resistome [6]. Household characteristics like building design, ventilation, and humidity can impact the indoor microbiome as well [7, 8]. Fu et al. (2020) found that ventilation, quality of the interior, and surface types impacted microbial community richness and composition [9]. Surface materials can also harbor specific types of microbes and can alter the indoor microbiome [10].
Specifically, moisture is essential for microbial growth in many settings and is common in approximately half of US homes [11, 12]. Thus, moisture, humidity, and mold growth are well known influences on the microbiome. Using 16s rRNA amplicon sequencing (16S) and ITS rRNA amplicon sequencing, previous studies have suggested that water leaks and mold presence promote bacterial and fungal richness and growth [5]. Lax et al. (2019) found that wetness drives microbial growth and wet materials can promote specific taxa within buildings [13]. Similarly, mold in the home was associated with microbial community variation in German homes [14].
Outdoor taxa can influence the indoor taxa, and geography largely drives differences in the microbiome across regions [9, 15, 16]. For example, fungi found inside the home often originate outside the home and may be dispersed through windows [15]. Similarly, outdoor contaminants can be brought indoors through shoes, clothing or pets [17]. Similarly, land use may impact the surrounding microbiome. Comparing traditional farms, industrial farms, and urban areas, Fu et al. (2021) noted differences in the house dust microbiome [18]. However, there is limited information characterizing the most common microbial communities pertaining to specific areas and landscapes. Additionally, the dominant environmental drivers may differ by geographic area. The microbiome of hotel rooms in Asia and Europe are distinctly shaped by proximity to sea and quality of the interior while microbial composition differed by road types and drinking water sources in homes along the Arizona-Sonora border [9, 16].
Previous publications using the Agricultural Lung Health Study (ALHS), a nested case control study of current asthma among farmers and their spouses in North Carolina (NC) and Iowa have investigated the house dust microbiome. In the ALHS, specific taxa were associated with pulmonary function and asthma, highlighting the impact of the house dust microbiome in a farming population [3, 19]. Additionally, the ALHS, as a farming population, provides a unique setting to investigate the house dust microbiome. For example, the ALHS measured exposures less common in the general population such as indoor wood burning. Similarly, the farm setting provides a distinct opportunity to gain further insight on the impact of the surrounding landscape, including proximity to crop and animal farming on the indoor microbiome.
The house dust microbiome has commonly been investigated using 16S rRNA sequencing, a technique with uncertainty at lower taxonomic ranks. This approach limits one’s ability to characterize the full range of microorganisms present in the microbiome. Whole metagenome shotgun sequencing (WGS) provides in-depth taxonomic classification resolution [20]. WGS can more accurately identify taxa at the species level and more comprehensively describe a microbial community. In the ALHS, the use of WGS has allowed for the expansion on 16S findings and the detection of novel microbial associations [3, 21]. Specifically, compared to 16S, Ziyue et al. (2023) found four times more unique microbial taxa using WGS [21]. Furthermore, the ALHS has a large sample size compared to other sequencing studies, providing greater power to identify distinct associations.
Here, we characterized metagenomic profiles in vacuumed house dust collected from 771 households in a US farming cohort. This extensive investigation enhances our understanding of indoor microbiome diversity, composition, and microbial correlations.
2. Methods
2.1. Study population and design
The ALHS is a case-control study of current asthma nested within the Agricultural Health Study (AHS), a prospective cohort study of licensed pesticide applicators, mostly farmers, and their spouses in Iowa and North Carolina [22]. A total of 3,301 ALHS participants were enrolled from 2009-2013 (data version P3REL201209.00): 1,223 asthma cases and 2,078 non-cases. Further details on the study design and inclusion criteria have previously been described [3, 21]. Written informed consent was obtained from all participants.
2.2. House dust collection and whole genome shotgun sequencing
Of the 3,301 participants, 2,871 received a home visit from a trained field technician. During the visit, the field technician vacuumed a one square yard (0.84-m2) area on the sleeping surface and on the floor next to the bed. These surfaces were vacuumed for 2 minutes each with a DUSTREAM Collector (Indoor Biotechnologies Inc., Charlottesville, VA, USA). Participants were asked to not clean the sampling area for 14 days prior to the home visit. Samples were collected across all seasons. A simple random sample of these participants (N=1000) was taken for 16S rRNA amplicon sequencing, of which 879 passed quality control. These same samples were later sent for whole genome shotgun sequencing.
Details on DNA extraction and processing have previously been described [21, 23]. Dust samples were sieved, weighed into aliquots of 50 mg, and frozen at -20 C at Social & Scientific Systems prior to DNA extraction. Library preparation, multiplexing, and whole genome shotgun metagenomic sequencing were performed at the University of California San Diego. Sequencing was conducted using an Illumina NovaSeq (Illumina, Inc., San Diego, CA, USA). The following novel methods were developed by Guccione et al. (2025) and were previously described in further detail [24]. Raw FASTQ files were filtered for low-quality reads, adapters, and duplicates using fastp (v0.23.4) [25]. A minimum cutoff length of 100 base pairs was used. Reads were filtered against host reads using the human reference genome GRCh38.p14 and then T2T-CHM13v2.0 using minimap2 (v2.28-r1209) and samtools (v1.12), respectively [26-29]. Afterwards, Movi (unversioned; git commit hash 76d5a6da1ec0aeb0121b5ac7c59b295936e23cc1) was used to filter against the 94 human reference genomes from the Human Pangenome Reference Consortium (HPRC) [30, 31]. Reads with pseudo-matching lengths were discarded using movi-default. Qiita was used to process the resulting sequences with default pipeline and parameters [32]. The Qiita study ID is 13186. Sequences were aligned to the Web of Life2 (WoL2) reference database via Woltka and using Bowtie2 [33-35]. Tables were generated using the Web of Life Toolkit App. Finally, 53 samples with sequencing depths less than 1000 and 5,733 taxa that appeared in less than 10% of samples were filtered out. Additionally, one sample was removed as an outlier identified in a NMDS plot. This resulted in a final 771 samples and 5,821 taxa for further analysis.
2.3. Household environmental exposures
Information on most household exposures was collected by a questionnaire. Additionally, during the home visit, the technician noted the presence of water leaks or moisture in different rooms in the home by observing the ceiling, windows, walls, plumbing, and floor. We created a binary variable describing presence of leaks overall in the home.
Our variables of interest were presence of leaks (Yes/No), presence of mold/mildew (Yes/No), year home was built in (1893-1940, 1941-1980, 1981-1996), frequency of AC use (Never, Moderate, Frequent), presence of an air filtration device (Yes/No), frequency of use of wood as heating fuel (Never, Secondary, Main), de-humidifier use (Yes/No), humidifier use (Yes/No), and removal of shoes before entering the home (Yes/No).
2.4. Microbial diversity analysis
We rarified data to the minimum library size (1,006) prior to calculating alpha and beta diversities. We used two alpha diversity measures, richness and Shannon index, to quantify intra-group diversity. All values were generated by the phyloseq R package (v1.44.0) [36]. To find associations between alpha diversity and household exposures, we fitted a linear regression model for each exposure with the diversity measure, richness or Shannon index, as the outcome. Shannon index was exponentially transformed. All models were adjusted for asthma (study selection factor). An additional model was also adjusted for current animal farming. We also conducted a stratified analysis by state of residence. The statistical significance threshold was set at p-value<0.05. Individuals with missing exposure information were excluded.
To evaluate inter-group diversity, we calculated two beta diversity measures: unweighted and weighted UniFrac distance metrics. We conducted permutational multivariate analysis of variance (PERMANOVA) to test the differences in microbial community structure across exposure groups [37]. The adonis method in the vegan R package (v2.6-8) reports the R2 value, which we use to quantify the percentage of variance explained [38]. Similar to the alpha diversity analysis, all models were adjusted for asthma. Additional analyses were also adjusted for current animal farming and stratified by state. To visualize differences by exposure level, we used principal coordinate analysis (PCoA) plots.
Supplemental analyses checking the influence of other covariates were also conducted. The first supplemental analysis adjusted for current animal farming and the second adjusted for state, home condition, and presence of cats or dogs.
2.5. Differential abundance analysis
To identify differentially abundant taxa by exposure levels, we used Analysis of Composition of Microbiomes with Bias Correction 2 (ANCOM-BC2, v2.2.2) models [39] with default options. ANCOMB-BC2 is based on a linear regression framework on log transformed taxa counts. ANCOM-BC2 accounts for variation in sequencing depth by estimating the sampling fraction. To apply a log transformation to taxa counts of zero, ANCOM-BC2 uses a pseudo-count and then conducts a sensitivity analysis to ensure the pseudo-count does not affect the results. The prevalence cutoff for taxa was set at 10%. We controlled the family-wise error rate (FWER) using the Holm-Bonferroni method [40]. We tested taxa at the species level and identified a taxon as significantly differentially abundant if it had a p-value<0.05 after the Holm-Bonferroni correction and if it passed the sensitivity analysis. Similar to the diversity analysis, all models were adjusted for the selection factor, asthma (Yes/No).
2.6. Microbial network analysis
We examined correlations between taxa in the microbiome using Sparse Estimation of Correlations among Microbiomes (SECOM, v2.2.2) [41]. We calculated a correlation matrix using Spearman correlation and the secom_linear function with default options (ANCOM-BC2, v2.2.2) [39]. SECOM accounts for the compositionality of microbiome data as well as sample-specific and taxon-specific biases. Microbial networks were created based on correlation coefficients from SECOM. To visualize results, we pruned taxa with a co-occurrence less than 10 and correlation coefficient less than 0.5 and constructed networks using the Network Construction and comparison for Microbiome data (NetCoMi) R package [42].
We performed all statistical analyses and visualizations in R v4.4.2 [43].
3. Results
3.1. Descriptive statistics
More participant’s homes were located in Iowa (68%) than North Carolina (32%) (Table 1). The most commonly reported exposure was frequent AC use (65%), followed by removal of shoes before entering the home (49%), de-humidifier use (42%), and humidifier use (35%). A leak was identified in 13% of households and mold/mildew was identified in 16% of households. De-humidifier use, humidifier use, frequency of AC use, frequency of air conditioning use, use of wood for heating, and home age showed some differences in prevalence by state.
Table 1: Study Population Characteristics.
| Exposure | Overall (N=771), N (%) | IA (N=528), N (%) | NC (N=243), N (%) | |
|---|---|---|---|---|
| Presence of leaks | No | 648 (87%) | 449 (88%) | 199 (84%) |
| Yes | 101 (13%) | 62 (12%) | 39 (16%) | |
| Missing | 22 | 17 | 5 | |
| Presence of mold/mildew | No | 644 (84%) | 435 (82%) | 209 (86%) |
| Yes | 127 (16%) | 93 (18%) | 34 (14%) | |
| Missing | 0 | 0 | 0 | |
| Frequency of air conditioning use | Never | 92 (12%) | 74 (14%) | 18 (7.4%) |
| Moderate | 179 (23%) | 150 (28%) | 29 (12%) | |
| Frequent | 499 (65%) | 303 (57%) | 196 (81%) | |
| Missing | 1 | 1 | 0 | |
| Presence of an air filtration device | No | 609 (80%) | 414 (79%) | 195 (81%) |
| Yes | 157 (20%) | 112 (21%) | 45 (19%) | |
| Missing | 5 | 2 | 3 | |
| Frequency of wood as heating fuel | Never | 613 (80%) | 445 (84%) | 168 (69%) |
| Secondary | 82 (11%) | 44 (8.3%) | 38 (16%) | |
| Main | 75 (9.7%) | 39 (7.4%) | 36 (15%) | |
| Missing | 1 | 0 | 1 | |
| De-humidifier use in the last 12 | No | 446 (58%) | 229 (43%) | 217 (89%) |
| Yes | 325 (42%) | 299 (57%) | 26 (11%) | |
| Missing | 0 | 0 | 0 | |
| Humidifier use in the last 12 | No | 499 (65%) | 291 (55%) | 208 (86%) |
| Yes | 272 (35%) | 237 (45%) | 35 (14%) | |
| Missing | 0 | 0 | 0 | |
| Removal of shoes before entering the home | No | 396 (51%) | 234 (44%) | 162 (67%) |
| Yes | 374 (49%) | 293 (56%) | 81 (33%) | |
| Missing | 1 | 1 | 0 | |
| Home Age | 1981-1996 | 143 (25%) | 82 (22%) | 61 (33%) |
| 1941-1980 | 222 (40%) | 118 (31%) | 104 (56%) | |
| 1893-1940 | 196 (35%) | 175 (47%) | 21 (11%) | |
| Missing | 210 | 153 | 57 | |
| Asthma status | Control | 480 (62%) | 321 (61%) | 159 (65%) |
| Case | 291 (38%) | 207 (39%) | 84 (35%) | |
| Missing | 0 | 0 | 0 | |
Our dust metagenomic profiles, after removing low quality samples and rare taxa, included 771 samples and 5,821 taxa. A total of 112,738,769 reads were present across all samples with an average of 146,224 reads per sample. In summary, among the taxa, 5,748 (99%) were described as bacteria and 73 (2%) as archaea. Forty-five unique phyla were identified with the most common being Proteobacteria (31%), Actinobacteriota (22%), Firmicutes (15%), Firmicutes_A (12%), and Bacteroidota (11%). Of the phyla identified in our study, 16 had an abundance greater than 1%. 2,250 unique genera were identified with the most common being Streptomyces (2%), Corynebacterium (2%), Prevotella (1%), Streptococcus (1%), and Mycobacterium (1%). Of the genera present, 34 had an abundance greater than 10%. Finally, 5,018 species were identified in our sample. The most common were Moraxella catarrhalis (0.7%), Lacticaseibacillus paracasei (0.3%), Limosilactobacillus fermentum (0.3%), Escheria coli (0.3%), and Methanosarcina mazei (0.3%).
3.2. Microbial diversity
We examined two overall microbial diversity measures (richness and Shannon index) within dust samples and found greater microbial diversity in homes reporting presence of mold/mildew compared to those reporting an absence (p-value<0.05) (Figure 1). Presence of leaks and removal of shoes before entering the home were also positively associated with both richness and Shannon index (p-value<0.1). Additionally, we found that humidifier use was positively associated with richness (p-value=0.09). Presence of leaks and mold/mildew remained positively associated with microbial diversity (p-value<0.01) when adjusting for current animal farming, but removal of shoes lost statistical significance (p-value=0.4). The direction of associations was concordant when stratified by state (Supplementary Figure S1).
Figure 1: Household Exposures and Alpha Diversity.
Effect size, representing the derived coefficient for each linear regression model with richness (A, C) or Shannon index (B, D) as the outcome. Figures A and B include each exposure in univariate analyses, without adjustment for other factors but with adjustment for asthma status, a selection factor. Figures C and D additionally adjust for current animal farming. The error bars represent 95% confidence intervals of the effect size. The reference group for all exposures is “no exposure” except for wood as heating fuel (vs never), frequency of air conditioning (AC) use (vs never), and years home built (vs 1981-1996).
Using weighted UniFrac distance, we found statistically significant differences in beta diversity for all exposures (p-value<0.05) (Supplementary Table S1, Supplementary Figure S2). The variance explained by the exposure levels were all relatively small, consistent with previous literature. Similar results were observed using the unweighted UniFrac distances (Supplementary Table S2). When adjusting for current animal farming, each exposure remained statistically significant with albeit still small, but notably larger R2 values (unweighted average R2 = 0.012, weighted average R2 = 0.016) (Supplementary Table S3). In analyses stratified by state, R2 values were slightly higher in NC (NC average R2=0.01, IA average R2=0.007) (Supplementary Table S4). These results support an impact, albeit modest, of each of these exposures on microbial composition.
3.3. Differential abundance
When we examined the 5,821 individual species for differential abundance, 640 species (from 427 genera across 26 phyla) were differentially abundant (p-value<0.05 after multiple testing correction) in relation to one or more exposures (Supplementary Table S5). Five exposures had at least 10 differentially abundant taxa and 7 exposures had less than 10 differentially abundant taxa (Figure 2, Figure 3, Supplementary Figure S3). The exposures with the largest number of differentially abundant taxa were de-humidifier use, humidifier use, and removal of shoes before entering the home (Figure 2). Notably, more differentially abundant taxa were identified for older homes (built in 1893-1996 vs 1981-1996), frequent AC use (vs never), and wood as the main heating fuel (vs never). Ninety eight percent of all significantly differentially abundant taxa were bacteria and 2% were archaea. As expected, the most common phyla among the differentially abundant taxa reflected the most common phyla in our study population: Proteobacteria (23%), Actinobacteriota (22%), Firmicutes_A (15%), Bacteroidota (14%), and Firmicutes (9%). The most common genera consisted of Prevotella (3%), Mycobacterium (3%), Bradyrhizobium (2%), Cryptobacteroides (2%), and Rickettsia (1%).
Figure 2: Differentially Abundant Taxa Related to Household Exposures (>=10 significant taxa).
Exposures with at least 10 significantly associated differentially abundant taxa are presented. Volcano plots of the significant differentially abundant taxa for frequency of AC use (frequent vs never) (A), wood as a heating fuel (main vs never) (B), de-humidifier use (C), humidifier use (D), removal of shoes before entering the home (E). Significant positively associated taxa are represented as red (p-value<0.05 after multiple testing correction, passed sensitivity analysis, LFC > 0), significant negatively associated taxa as blue (p-value<0.05 after multiple testing correction, passed sensitivity analysis, LFC < 0), and non-significant taxa as gray (p-value>0.05 after multiple testing correction, did not pass sensitivity analysis). The horizontal dash line represents the threshold for the p-value after the Holm-Bonferroni multiple testing correction. The reference group for all exposures is “no exposure” except for wood as heating fuel (vs never), frequency of air conditioning (AC) use (vs never), and years home built (vs 1981-1996).
Figure 3: Differentially Abundant Taxa Related to Household Exposures (<10 significant taxa).
Exposures with less than 10 significantly associated differentially abundant taxa are presented. Log fold change (LFC) is plotted for the significantly differentially abundant taxa related to presence of leaks (A), presence of mold/mildew (B), frequency of AC use (moderate vs never) (C), wood as heating fuel (secondary vs never) (D), presence of an air filtration device (E), home built in 1893-1940 (vs 1961-1996) (F), home built in 1941-1980 (vs 1981-1996) (G). Significant positively associated taxa are represented as red (p-value<0.05 after multiple testing correction, passed sensitivity analysis, LFC > 0) and significant negatively associated taxa as cyan (p-value<0.05 after multiple testing correction, passed sensitivity analysis, LFC < 0).
De-humidifier use was associated with 431 differentially abundant taxa, 206 with an increased abundance and 225 with a decreased abundance. Among the most common genera associated with de-humidifier use were taxa from Mycobacterium, Bradyrhizobium, and Methylobacterium. Additionally, Streptomyces and Paraburkholderia were negatively associated with de-humidifier use. Humidifier use was associated with 174 differentially abundant taxa, 73 with an increased abundance and 101 with a decreased abundance. Of those taxa with a decreased abundance, many were in the genera Bradyrhizobium, Paraburkholderia, and Terriglobus. Additionally, two of the top differentially abundant taxa (ranked by p-value after multiple testing correction) were members of the genera Terriglobus and Caballeronia. Removal of shoes before entering the home was associated with 196 differentially abundant taxa. Of those taxa, 135 had an increased abundance and 61 had a decreased abundance. The genera Prevotella, Cryptobacteroides, and Clostridium were commonly positively associated with removal of shoes. Similarly, many of the top differentially abundant taxa were a member of these genera as well as Clostridium, Glaesserella, and Treponema. Additionally, taxa from the genera Bradyrhizobium, Paraburkholderia, and Caballeronia were negatively associated with removal of shoes. Finally, wood as the main heating fuel was associated with 24 differentially abundant taxa, 17 with an increased abundance and 7 with a decreased abundance. Two of the most commonly associated genera, Streptomyces and Mucilaginibacter, were positively associated with wood as the main heating fuel. Additionally, among the top differentially abundant taxa, we observed the genera Glaesserella, Paraburkholderia, and Treponema_D. The majority of differentially abundant taxa (69% of 620 species) were associated with one exposure tested (Supplementary Figure S5).
Our differential abundance analysis was run at the species level, resulting in the discovery of a large number of novel associations of species related to household exposures (Table 2). Several of these species serve as notable indicators for specific exposures. Leifsonia cynodontis and Terriglobus roseus_B were negatively associated with humidifier use. Leifsonia cynodontis was also negatively associated with de-humidifier use alongside Bacillus_AG koreensis and Leptolyngbya boryana. Finally, Glaesserella parasuis and Treponema_D berlinense were positively associated with removal of shoes before entering the home.
Table 2: Top 5 differentially abundant taxa in house dust by exposure.
| Taxonomic classification | LFC | FWER- adjusted P- value |
Primary Environment/ Isolation Location |
||||
|---|---|---|---|---|---|---|---|
| Kingdom | Phylum | Family | Genus | Species | |||
| Presence of leaks | |||||||
| Bacteria | Actinobacteriota | Coriobacteriaceae | Enorma | Enorma massiliensis | 0.738 | 3.14E-03 | Human fecal flora |
| Bacteria | Cyanobacteria | Leptolyngbyaceae | Leptolyngbya | Leptolyngbya sp001548435 | −0.6145 | 3.72E-03 | Marine environments |
| Bacteria | Firmicutes_A | Lachnospiraceae | C-53 | C-53 sp003612525 | 0.6852 | 5.56E-03 | |
| Bacteria | Methylomirabilota | CSP1-6 | AR12 | AR12 sp003220225 | 0.5955 | 1.15E-02 | |
| Bacteria | Firmicutes_A | T1SED10-28 | Isachenkonia | Isachenkonia alkalipeptolytica | −0.555 | 2.42E-02 | Alkaline lake sediment |
| Presence of mold/mildew | |||||||
| Bacteria | Bacteroidota | Chitinophagaceae | Phnomibacter | Phnomibacter ginsenosidimutans | −0.7568 | 7.22E-08 | Sewage wastewater |
| Bacteria | Firmicutes_A | Lachnospiraceae | Extibacter | Extibacter hylemonae | −0.6718 | 1.66E-06 | Gut microbiome |
| Bacteria | Proteobacteria | Azospirillaceae | Niveispirillum | Niveispirillum sp001296005 | −0.6387 | 3.06E-05 | |
| Bacteria | Actinobacteriota | Eggerthellaceae | CAG-1427 | CAG-1427 sp000435475 | −0.5974 | 1.60E-04 | |
| Bacteria | Firmicutes | Lactobacillaceae | Ligilactobacillus | Ligilactobacillus ruminis | 0.6963 | 1.18E-02 | Mammalian gut microbiome |
| Frequency air conditioning use: moderate vs never | |||||||
| Bacteria | Fusobacteriota | Leptotrichiaceae | Sneathia | Sneathia vaginalis | 0.9929 | 5.19E-03 | Genital microbiome |
| Frequency air conditioning use: frequent vs never | |||||||
| Bacteria | Cyanobacteria | Nostocaceae | 1.4065 | 9.08E-14 | |||
| Archaea | Thermoproteota | Nitrososphaeraceae | UBA10452 | UBA10452 sp009898475 | 1.4726 | 5.32E-13 | |
| Bacteria | Cyanobacteria | Chroococcidiopsidaceae | Chroococcidiopsis | Chroococcidiopsis thermalis | 1.1254 | 1.48E-08 | Extreme environments |
| Bacteria | Actinobacteriota | Microbacteriaceae | Canibacter | Canibacter oris | 1.142 | 1.60E-08 | Canine oral microbiome |
| Bacteria | Actinobacteriota | Mycobacteriaceae | Nocardia | Nocardia mikamii | 1.1159 | 1.93E-08 | Human pulmonary infection |
| Presence of air filtration device | |||||||
| Archaea | Halobacteriota | Methanosarcinaceae | Methanosarcina | Methanosarcina thermophila | 0.6512 | 3.12E-08 | Biogas fermenter |
| Bacteria | Firmicutes_A | Lachnospiraceae | Butyribacter | Butyribacter sp003149875 | 0.6476 | 4.77E-07 | |
| Bacteria | Proteobacteria | Rhodocyclaceae | Azonexus | Azonexus hydrophilus | 0.4721 | 3.18E-04 | Freshwater |
| Bacteria | Actinobacteriota | Rubrobacteraceae | Rubrobacter_A | Rubrobacter_A aplysinae | 0.4495 | 2.78E-03 | Marine sponge |
| Bacteria | Actinobacteriota | Bifidobacteriaceae | Scardovia | Scardovia wiggsiae | 0.5694 | 5.16E-03 | Caries |
| Wood as heating fuel: secondary vs never | |||||||
| Bacteria | Proteobacteria | Acetobacteraceae | Acidisoma | Acidisoma sp009765865 | 1.1656 | 2.90E-06 | |
| Bacteria | Proteobacteria | Beijerinckiaceae | Methylovirgula | Methylovirgula ligni | 0.9243 | 8.93E-04 | Decaying wood |
| Bacteria | Proteobacteria | Burkholderiaceae | Caballeronia | Caballeronia udeis | 0.8118 | 1.21E-02 | Soil microbiome |
| Bacteria | Firmicutes_A | Lachnospiraceae | CAG-882 | CAG-882 sp000435595 | −0.7102 | 2.94E-02 | |
| Bacteria | Proteobacteria | Burkholderiaceae | Paraburkholderia | Paraburkholderia xenovorans | 0.7418 | 3.67E-02 | Soil microbiome |
| Wood as heating fuel: main vs never | |||||||
| Bacteria | Proteobacteria | Pasteurellaceae | Glaesserella | Glaesserella parasuis | −1.2369 | 8.08E-07 | Swine pathogen |
| Bacteria | Proteobacteria | Burkholderiaceae | Paraburkholderia | Paraburkholderia xenovorans | 1.1521 | 4.91E-06 | Soil microbiome |
| Bacteria | Proteobacteria | Acetobacteraceae | Acidisoma | Acidisoma sp009765865 | 1.1147 | 3.91E-04 | |
| Bacteria | Firmicutes_C | Selenomonadaceae | Selenomonas_C | Selenomonas_C bovis | −1 | 4.03E-04 | Yak rumen microbiome |
| Bacteria | Firmicutes | Erysipelotrichaceae | Floccifex | Floccifex porci | −1.0054 | 6.79E-04 | Swine intestinal microbiome |
| De-humidifier use | |||||||
| Bacteria | Actinobacteriota | Microbacteriaceae | Leifsonia | Leifsonia cynodontis | −1.0643 | 3.20E-39 | Grass pathogen |
| Bacteria | Cyanobacteria | Nostocaceae | Anabaenopsis | Anabaenopsis sp000316625 | −0.908 | 1.30E-37 | |
| Bacteria | Proteobacteria | Burkholderiaceae | Caballeronia | Caballeronia zhejiangensis | −1.2257 | 4.11E-37 | Wastewater treatment |
| Bacteria | Cyanobacteria | Chroococcidiopsidaceae | Chroococcidiopsis | Chroococcidiopsis thermalis | −1.2051 | 2.91E-35 | Extreme environments |
| Bacteria | Cyanobacteria | Nostocaceae | Fischerella | Fischerella prolifica | −0.8628 | 3.88E-33 | |
| Humidifier use | |||||||
| Bacteria | Firmicutes | Bacillaceae_H | Bacillus_AG | Bacillus_AG koreensis | −1.0676 | 1.28E-34 | Rhizosphere soil microbiome |
| Bacteria | Bacteroidota | Sphingobacteriaceae | Mucilaginibacter | Mucilaginibacter corticis | −0.7881 | 1.91E-24 | Tree bark |
| Archaea | Thermoproteota | Nitrososphaeraceae | UBA10452 | UBA10452 sp009898475 | −0.7479 | 8.45E-22 | |
| Bacteria | Proteobacteria | Burkholderiaceae | Caballeronia | Caballeronia zhejiangensis | −0.9097 | 4.71E-19 | Wastewater treatment |
| Bacteria | Actinobacteriota | Streptosporangiaceae | Bog-532 | Bog-532 sp003164955 | −0.6928 | 6.71E-19 | |
| Removal of shoes before entering the home | |||||||
| Bacteria | Firmicutes | UBA660 | CAG-533 | CAG-533 sp000434495 | 0.8011 | 7.44E-25 | |
| Bacteria | Proteobacteria | Rhodocyclaceae | SFHR01 | SFHR01 sp004555545 | 0.7695 | 1.72E-21 | |
| Bacteria | Bacteroidota | UBA932 | Cryptobacteroides | Cryptobacteroides sp900316045 | 0.6979 | 2.22E-20 | Human gut microbiome |
| Bacteria | Proteobacteria | Pasteurellaceae | Glaesserella | Glaesserella parasuis | 0.7544 | 2.07E-18 | Swine pathogen |
| Bacteria | Bacteroidota | F082 | Limimorpha | Limimorpha sp900318085 | 0.5444 | 2.14E-17 | Ruman microbiome |
| Home built in 1893-1940 (vs 1981-1996) | |||||||
| Bacteria | Cyanobacteria | Chroococcidiopsidaceae | Chroococcidiopsis | Chroococcidiopsis thermalis | −1.3284 | 4.89E-07 | Extreme environments |
| Bacteria | Cyanobacteria | Nostocaceae | −1.2335 | 6.32E-07 | |||
| Bacteria | Actinobacteriota | Mycobacteriaceae | Corynebacterium | Corynebacterium freiburgense | −1.1098 | 4.10E-05 | Canine oral microbiome |
| Bacteria | Actinobacteriota | Antricoccaceae | Antricoccus | Antricoccus suffuscus | −0.9225 | 1.12E-03 | Cave soil |
| Bacteria | Actinobacteriota | Streptosporangiaceae | Bog-532 | Bog-532 sp003164955 | −0.835 | 4.81E-03 | |
| Home built in 1941-1980 (vs 1981-1996) | |||||||
| Bacteria | Firmicutes | Erysipelotrichaceae | Bulleidia | Bulleidia sp902793635 | −0.7803 | 4.60E-03 | |
| Bacteria | Cyanobacteria | Gastranaerophilaceae | Stercorousia | Stercorousia sp001765415 | −0.808 | 7.53E-03 | |
| Bacteria | Bacteroidota | P3 | UBA10566 | UBA10566 sp002399025 | −0.7643 | 2.51E-02 | |
3.4. Microbial networks
Microbial networks were constructed for presence of leaks, presence of mold/mildew, de-humidifier use, and humidifier use. These variables were chosen as they are all moisture-related and relevant to microbial diversity and differential abundance in our study. We found some interactions present in one network and absent in another (Figure 4). More interactions were observed among homes exposed to leaks than those not exposed to leaks. For example, the phylum Planctomycetota was correlated with Gemmatimonadota and Deinococcota in homes exposed to leaks. Similarly, a link was observed between Proteobacteria and Cyanobacteria in homes without de-humidifier use while conversely, Firmicutes_A was linked to Verrucomicrobiota in homes with de-humidifier use. Notably, the vast majority of interactions were positive, and interactions were observed across kingdoms. The phyla Firmicutes_A, Firmicutes_C, and Bacteroidota were strongly linked together across all exposures. Additionally, the phyla Camplyobacterota, Deinococcota, and Marinisomatota appear across several microbial networks but do not contain significant differentially abundant taxa. The presence of each condition was also accompanied by the inclusion of the phylum Patescibacteria in the microbial networks.
Figure 4: Microbial networks by Moisture-Based Environmental Exposures.
Microbial networks constructed based on Spearman’s correlation using SECOM calculated for presence of leaks (A), presence of mold/mildew (B), no de-humidifier use (C), humidifier use (D). The color of circles is based on kingdom. The width of connecting lines is based on the magnitude of the correlation coefficient. Interactions unique to that exposure (yes compared to no) are colored purple. Nodes are sized based on abundance. To visualize results, we pruned taxa with a co-occurrence less than 10 and correlation coefficient less than 0.5.
4. Discussion
In this large whole metagenome sequencing study, we identified household environmental exposures influencing the house dust microbiome. In particular, moisture-related exposures (presence of mold/mildew, presence of leaks, humidifier use, and de-humidifier use) appear to influence the diversity and composition of the microbiome. Additionally, removal of shoes was linked with the introduction of outdoor taxa to the indoor microbiome. These results are strengthened through the use of additional host filtering during WGS processing and a modern reference genome database.
Our findings regarding moisture-related home conditions are in line with previous reports. In our study, presence of mold/mildew was positively associated with overall microbial diversity. Additionally, presence of leaks and humidifier use were borderline positively associated with richness (p-value<0.1). Previously, bacterial and fungal richness have been demonstrated to be associated with water leaks using 16S sequencing (n=196) [5]. Similarly, a separate study (n=35), showed greater fungal diversity was seen in homes with high moldiness values, water damage, and humidity [44].
Moisture in the home also appeared to impact the microbial composition of the house dust microbiome. Specifically, de-humidifier use was negatively associated with the genera Streptomyces and Mycobacterium, of which are well known soil inhabitants and have previously been linked to moisture in homes [45]. Streptomyces and Mycobacterium have been commonly isolated in water-damaged buildings and materials [46, 47]. Similarly, higher concentrations of these taxa were associated with visible mold [48]. Hence, it would be expected to see a decreased abundance of moisture-related taxa like Streptomyces and Mycobacterium in response to de-humidifier use. Additionally, the presence of moisture appeared to alter microbial interactions in the home. Microbial networks for homes with leaks and humidifier use had more connections than homes without those exposures, suggesting that moisture can facilitate more complex microbial communities. This increase in complexity could be in part due to greater microbial diversity associated with moisture. Unique interactions were also observed across exposures. Building upon these findings, moisture related exposures may result in unique microbial interplay in the house dust microbiome.
Using metagenomic sequencing, we also identified many novel associations at the species level. Among the top differentially abundant taxa associated with humidifier use, we identified Terriglobus roseus_B and Leifsonia cynodontis, two taxa commonly found in the environment. Terriglobus roseus_B is a ubiquitous soil bacterium and Leifsonia cynodontis is a vascular pathogen of Bermuda grasses [49, 50]. Similarly, the environmental and plant associated genera Bradyrhizobium, Paraburkholderia, and Terriglobus were negatively associated with de-humidifier use and humidifier use [50-52]. Specifically, these genera have important parts of the rhizosphere and agricultural soil [45, 50]. Considering that taxa with outdoor sources are known to be found within the indoor microbiome, it is notable that these taxa have a decreased abundance in relation to the use of humidity-related devices [15]. As such, it is possible that humidifier use or de-humidifier use somehow selects away from outdoor taxa. For example, individuals using a humidifier may keep their windows closed, limiting outdoor taxa from entering the home.
Occupant hygiene habits are less frequently studied in relation to the household microbiome. Shoe soles are known to transport microbes but to the best of our knowledge, removal of shoes before entering the home has not previously been measured in a house dust microbiome study [53]. Removal of shoes before entering the home was negatively associated with the environmental taxa Bradyrhizobium, Paraburkholderia, and Terriglobus [50-52]. This decreased abundance suggests that these taxa can be introduced through an occupant’s shoes and their removal prevents their entry. Notably, removal of shoes was also positively associated with microbial diversity inside homes. This is also reflected in the differentially abundant taxa positively associated with removal of shoes before entering the home. Among both the most common and top differentially abundant taxa, we identified genera often isolated from animal’s microbiomes, specifically Clostridium, Cryptobacteroides, and Prevotella [4, 54, 55]. Additionally, many species positively associated with removal of shoes before entering the home are connected to farming. For example, Glaesserella parasuis is an infectious disease of pigs and Treponema_D berlinense has been isolated from porcine feces [56, 57]. It has been established that the outdoor environment exerts an influence on the indoor microbiome [15]. Upon first glance, these results contradict previous findings and the exact reason for this directionality is unclear. However, because this is a farming population, these taxa would be expected to be seen in the surrounding landscape and thereby possibly the indoor microbiome. Taken together, removal of shoes could act as proxy for hygiene. Hence, those who choose to remove their shoes, may take part in work with more exposures to diverse microbial sources and bring more outdoor taxa inside on clothing or skin. This is further confirmed by adjusting for current animal farming in the microbial diversity models (Figure 1C, 1D). Notably, significance of the association between richness and removal of shoes is lost after adjustment, suggesting that the relationship between removal of shoes and microbial diversity is dependent on animal farming. As such, if a participant is not currently farming, there would be few to no contaminants on their shoes and their removal would have no effect on the house dust microbiome. Additionally, adjustment for current animal farming does not noticeably alter the model results for any other exposure besides richness and humidifier use, suggesting that the impact of current animal farming is relatively unique to removal of shoes. With this reasoning, the removal of shoes is likely not the underlying metric measured and instead represents a novel determinant related to the introduction of outdoor taxa. Previously, the gut microbiome of swine farmers has been shown to differ from those without farming exposures [58]. Additionally, waste workers clothes are known to transport microorganisms from their work environment [59]. It is reasonable that farmwork could then also alter the microbiome of both the body and clothing of farmworkers, allowing those participants to transport farming taxa into the home. This provides an opportunity for further investigation into personal and home hygiene habits (i.e. washing hands, contamination of clothing, surface cleaning) that may reveal further mechanistic information in relation to microbiome diversity.
Several household exposures have previously been found to influence both microbial diversity and composition in the ALHS [21]. Specifically, presence of indoor pets was positively associated with microbial diversity and greater home cleanliness was negatively associated with microbial diversity [21]. Many farming-specific exposures were linked to both microbial diversity and composition [21]. In this investigation, we analyzed household environmental exposures which had not been thoroughly interrogated and found significant associations between the house dust metagenome, moisture-based exposures, and outdoor-related exposures. We also confirmed that an adjustment for previously examined exposures did not substantially alter the associations (Supplementary Table S6, S7). Of note, given that the ALHS is a farming population, this study provides an opportunity to gain a greater understanding of how the surrounding outdoor environment influences the microbiome. Specifically, it is known that occupation and the surrounding landscape can influence the house dust microbiome [9, 60]. In particular, farm-, plant-, and livestock- related microbes served as a good indicator for outdoor taxa, of which were associated with removal of shoes before entering the home.
One limitation of this study is that dust was only collected from the bedroom, limiting the generalizability to the rest of the home. Samples from multiple locations in the home could provide further insight into the entry of outdoor taxa and the impact of various housing exposures. Nevertheless, humans spend about a third of their time within the bedroom. Thus, the bedroom is an informative single sampling location. We also collected samples only at a single point in time. Thus, we had a limited snapshot of the indoor microbiome. This would tend to decrease our ability to find associations. Additionally, the generalizability of our findings is limited by this being a farming population. Urbanicity is known to impact the house dust microbiome, and with a primarily rural population we lose insight into this relationship [61]. However, rurality and farming exposures are known to increase microbial diversity, possibly providing a microbial population with greater depth [61]. Unfortunately, information on fungi was not available in our reference genome database when processing the samples. As such, this leaves a gap in our description of the house dust microbiome. Fungi such as Aspergillus are known to be common members of the indoor microbiome and future studies should expand the reference genome database to include the fungal kingdom [4]. Nevertheless, while we do not have an independent house dust microbiome dataset to replicate our findings, we can validate our findings through similarities with published data. The most common phyla in our sample were Proteobacteria and Actinobacteria, both of which are well established in the house dust microbiome. Proteobacteria is the largest phyla of bacteria and Proteobacteria and Actinobacteria are commonly found in the house dust microbiome [7, 8]. Similarly, the make-up of genera in our sample reflects previous house dust microbiome studies. For example, it is known that one of the strongest influences on the indoor microbiome is occupancy, specifically its human occupants. Acinetobacter, Staphylococcus, and Corynebacterium were among our most common genera and are well-known human commensals, commonly found in house dust [4, 7, 8].
Our study has several strengths. The use of whole genome sequencing (WGS) provided broad benefits to this study. Compared to 16S rRNA sequencing, WGS directly sequences fragments of the genome, improving accuracy. Of note, we identified differentially abundant taxa at the species level, which has been rarely reported on in relation to the house dust microbiome. Thus, we identified many novel species in relation to multiple household exposures. We also found the phylum Patescibacteria, a diverse clade of bacteria but have been commonly underreported in 16S sequencing [62]. This finding is notable as the use of WGS was likely necessary to capture this phylum’s presence. Our study also helped to characterize the house dust microbiome specifically in the context of unique farming cohort. Both occupant behavior and location are known to influence the indoor microbiome [6, 9]. This study can help elucidate how these exposures may differ or remain the same compared to other environments. Additionally, the sample size was a key strength of this study, providing greater statistical power than many previous studies. We also employed the novel method SECOM which helped to provide insight into microbial networks, and the interplay between microbes and composition of the house dust microbiome.
The greatest strength of this study is the utilization of novel host filtration methods developed by Guccione et al. (2025) and a state-of-the-art reference genome data base, Web of Life2 (WoL2) [24, 33]. While most host filtration tools only use one human reference, our data was filtered against two human reference genomes and the human pangenome reference consortium (HPRC), drastically decreasing the possibility of mis-mapped taxa. Additionally, WoL2 includes over 10,000 genomes and covers a broad array of taxa. As such, our data includes both well-known and novel taxa, many of which have yet to be thoroughly investigated. For example, presence of leaks was associated with Leptolyngbya sp001548435, C-53 sp003612525, and AR12 sp003220225. To the best of our knowledge, these taxa have not previously appeared in any literature. Our study provides a starting point for further research into many new taxa.
5. Conclusion
This study described the impact of household environmental exposures on the indoor house dust microbiome. Specifically, moisture and the surrounding landscape exerted the greatest influence on both microbiome diversity and composition. These findings helped to elucidate the role household exposures on the house dust microbiome and its influence on human health. Additionally, through the use of updated WGS methodology, our study identified many novel taxa associated with these exposures. Future studies may use these findings to identify and test possible intervention targets.
Supplementary Material
See Supplemental Materials Table of Contents document for list of tables and figures referenced.
Highlights:
Moisture and the surrounding landscape influence the house dust microbiome
Whole metagenome sequencing reveals novel taxa in relation to household exposures
Microbial network analysis reveals microbial interconnections
Acknowledgement:
We thank Dr. G Ackerman and Dr. G Humphrey for laboratory processing, and Dr. F Day and Dr. M Richard of Westat Inc for computational assistance.
This research was supported in part by the Intramural Research Program of the National Institutes of Health (NIH). The contributions of the NIH author(s) were made as part of their official duties as NIH federal employees, are in compliance with agency policy requirements, and are considered Works of the United States Government. However, the findings and conclusions presented in this paper are those of the author(s) and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services.
Funding information:
This study was supported by the Intramural Research Program of the National Institutes of Health (NIH), the National Institute of Environmental Health Sciences (Z01-ES049030 and Z01-ES102385, Z01- ES103390) and the National Cancer Institute (Z01-CP010119B), and by American Recovery and Reinvestment Act funds.
Footnotes
COI:
All authors declare no commercial, financial, or other competing interests that would lead to a conflict of interest for this paper.
Ethics:
The Institutional Review Board at the National Institute of Environmental Health Sciences approved this study. Written informed consent was obtained from all participants.
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Data Sharing:
Metagenome data used for this analysis are available at the Sequence Read Archive (SRA) under project number PRJNA975673 (https://www.ncbi.nlm.nih.gov/sra/).
References
- 1.Ege MJ, et al. , Exposure to environmental microorganisms and childhood asthma. N Engl J Med, 2011. 364(8): p. 701–9. [DOI] [PubMed] [Google Scholar]
- 2.Rook GA, Regulation of the immune system by biodiversity from the natural environment: an ecosystem service essential to health. Proc Natl Acad Sci U S A, 2013. 110(46): p. 18360–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lee M, et al. , House dust metagenome and pulmonary function in a US farming population. Microbiome, 2024. 12(1): p. 129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Barberán A, et al. , The ecology of microscopic life in household dust. Proc Biol Sci, 2015. 282(1814). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Dannemiller KC, et al. , Influence of housing characteristics on bacterial and fungal communities in homes of asthmatic children. Indoor Air, 2016. 26(2): p. 179–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Mahnert A, et al. , Man-made microbial resistances in built environments. Nature Communications, 2019. 10(1): p. 968. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kembel SW, et al. , Architectural design drives the biogeography of indoor bacterial communities. PLoS One, 2014. 9(1): p. e87093. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Rai S, Singh DK, and Kumar A, Microbial, environmental and anthropogenic factors influencing the indoor microbiome of the built environment. Journal of Basic Microbiology, 2021. 61(4): p. 267–292. [DOI] [PubMed] [Google Scholar]
- 9.Fu X, et al. , Continental-Scale Microbiome Study Reveals Different Environmental Characteristics Determining Microbial Richness, Composition, and Quantity in Hotel Rooms. mSystems, 2020. 5(3): p. 10.1128/msystems.00119-20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sun Y, et al. , Shotgun metagenomics of dust microbiome from flight deck and cabin in civil aviation aircraft. Indoor Air, 2020. 30(6): p. 1199–1212. [DOI] [PubMed] [Google Scholar]
- 11.Hoeksma P, Aarnink AJA, and Ogink N, Effect of temperature and relative humidity on the survival of airborne bacteria. 2015: Wageningen UR: Livestock Research Wageningen. [Google Scholar]
- 12.Group LIE, IAQ Report - Prevalence of Building Dampness. 2025. [Google Scholar]
- 13.Lax S, et al. , Microbial and metabolic succession on common building materials under high humidity conditions. Nature Communications, 2019. 10(1): p. 1767. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Weikl F, et al. , Fungal and Bacterial Communities in Indoor Dust Follow Different Environmental Determinants. PLoS One, 2016. 11(4): p. e0154131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Adams RI, et al. , Airborne bacterial communities in residences: similarities and differences with fungi. PLoS One, 2014. 9(3): p. e91283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Benton LD, et al. , Environmental and structural factors associated with bacterial diversity in household dust across the Arizona-Sonora border. Sci Rep, 2024. 14(1): p. 12803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nishioka MG, et al. , Distribution of 2,4-Dichlorophenoxyacetic Acid in Floor Dust throughout Homes Following Homeowner and Commercial Lawn Applications: Quantitative Effects of Children, Pets, and Shoes. ACS Publications, 1999. 33(9): p. 1359–1365. [Google Scholar]
- 18.Fu X, Ou Z, and Sun Y, Indoor microbiome and allergic diseases: From theoretical advances to prevention strategies. Eco-Environment & Health, 2022. 1(3): p. 133–146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lee MK, et al. , House dust microbiota in relation to adult asthma and atopy in a US farming population. J Allergy Clin Immunol, 2021. 147(3): p. 910–920. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rausch P, et al. , Comparative analysis of amplicon and metagenomic sequencing methods reveals key features in the evolution of animal metaorganisms. Microbiome, 2019. 7(1): p. 133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wang Z, et al. , Metagenomics reveals novel microbial signatures of farm exposures in house dust. Front Microbiol, 2023. 14: p. 1202194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Alavanja MC, et al. , The Agricultural Health Study. Environ Health Perspect, 1996. 104(4): p. 362–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Lee MK, et al. , Exposures Related to House Dust Microbiota in a U.S. Farming Population. Environ Health Perspect, 2018. 126(6): p. 067001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Guccione C, et al. , Incomplete human reference genomes can drive false sex biases and expose patient-identifying information in metagenomic data. Nature Communications, 2025. 16(1): p. 825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Chen S, et al. , fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics, 2018. 34(17): p. i884–i890. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Li H, Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics, 2018. 34(18): p. 3094–3100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Li H, et al. , The Sequence Alignment/Map format and SAMtools. Bioinformatics, 2009. 25(16): p. 2078–2079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Schneider VA, et al. , Evaluation of GRCh38 and de novo haploid genome assemblies demonstrates the enduring quality of the reference assembly. Genome Res, 2017. 27(5): p. 849–864. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Nurk S, et al. , The complete sequence of a human genome. Science, 2022. 376(6588): p. 44–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Zakeri M, et al. , Movi: a fast and cache-efficient full-text pangenome index. bioRxiv, 2024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Liao W-W, et al. , A draft human pangenome reference. Nature, 2023. 617(7960): p. 312–324. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Gonzalez A, et al. , Qiita: rapid, web-enabled microbiome meta-analysis. Nat Methods, 2018. 15(10): p. 796–798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Zhu Q, et al. , Phylogenomics of 10,575 genomes reveals evolutionary proximity between domains Bacteria and Archaea. Nat Commun, 2019. 10(1): p. 5477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Zhu Q, et al. , Phylogeny-Aware Analysis of Metagenome Community Ecology Based on Matched Reference Genomes while Bypassing Taxonomy. mSystems, 2022. 7(2): p. e0016722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Langmead B and Salzberg SL, Fast gapped-read alignment with Bowtie 2. Nat Methods, 2012. 9(4): p. 357–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.McMurdie PJ and Holmes S, phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One, 2013. 8(4): p. e61217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Anderson MJ, Permutational Multivariate Analysis of Variance (PERMANOVA), in Wiley StatsRef: Statistics Reference Online. p. 1–15. [Google Scholar]
- 38.Oksanen J, et al. , Vegan: Community Ecology Package. 2024. [Google Scholar]
- 39.Lin H and Peddada SD, Multigroup analysis of compositions of microbiomes with covariate adjustments and repeated measures. Nature Methods, 2024. 21(1): p. 83–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Holm S, A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 1979. 6(2): p. 65–70. [Google Scholar]
- 41.Lin H, Eggesbø M, and Peddada SD, Linear and nonlinear correlation estimators unveil undescribed taxa interactions in microbiome data. Nature Communications, 2022. 13(1): p. 4946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Peschel S, et al. , NetCoMi: network construction and comparison for microbiome data in R. Brief Bioinform, 2021. 22(4). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Team RC, R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, 2024. [Google Scholar]
- 44.Kettleson EM, et al. , Key determinants of the fungal and bacterial microbiomes in homes. Environ Res, 2015. 138: p. 130–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Loos D, et al. , A global survey of host, aquatic, and soil microbiomes reveals shared abundance and genomic features between bacterial and fungal generalists. Cell Reports, 2024. 43(4): p. 114046. [DOI] [PubMed] [Google Scholar]
- 46.Torvinen E, et al. , Mycobacteria and fungi in moisture-damaged building materials. Appl Environ Microbiol, 2006. 72(10): p. 6822–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Rintala H, Nevalainen A, and Suutari M, Diversity of streptomycetes in water-damaged building materials based on 16S rDNA sequences. Letters in Applied Microbiology, 2002. 34(6): p. 439–443. [DOI] [PubMed] [Google Scholar]
- 48.Kettleson E, et al. , Stenotrophomonas, Mycobacterium, and Streptomyces in home dust and air: associations with moldiness and other home/family characteristics. Indoor Air, 2013. 23(5): p. 387–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Monteiro-Vitorello CB, et al. , Complete Genome Sequence of Leifsonia xyli subsp. cynodontis Strain DSM46306, a Gram-Positive Bacterial Pathogen of Grasses. Genome Announc, 2013. 1(6). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Eichorst SA, Breznak JA, and Schmidt TM, Isolation and Characterization of Soil Bacteria That Define Terriglobus gen. nov., in the Phylum Acidobacteria. Applied and Environmental Microbiology, 2007. 73(8): p. 2708–2717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Dobritsa AP and Samadpour M, Transfer of eleven species of the genus Burkholderia to the genus Paraburkholderia and proposal of Caballeronia gen. nov. to accommodate twelve species of the genera Burkholderia and Paraburkholderia. Int J Syst Evol Microbiol, 2016. 66(8): p. 2836–2846. [DOI] [PubMed] [Google Scholar]
- 52.VanInsberghe D, et al. , Non-symbiotic Bradyrhizobium ecotypes dominate North American forest soils. Isme j, 2015. 9(11): p. 2435–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Rashid T, et al. , Shoe soles as a potential vector for pathogen transmission: a systematic review. Journal of Applied Microbiology, 2016. 121(5): p. 1223–1231. [DOI] [PubMed] [Google Scholar]
- 54.Lin L, et al. , Lignocellulolytic microbiomes orchestrating degradation cascades in the rumen of dairy cattle and their diet-influenced key degradation phases. Animal Advances, 2024. 1(1). [Google Scholar]
- 55.Holman DB, Gzyl KE, and Kommadath A, The gut microbiome and resistome of conventionally vs. pasture-raised pigs. Microbial Genomics, 2023. 9(7). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Nordhoff M, et al. , Treponema berlinense sp. nov. and Treponema porcinum sp. nov., novel spirochaetes isolated from porcine faeces. Int J Syst Evol Microbiol, 2005. 55(Pt 4): p. 1675–1680. [DOI] [PubMed] [Google Scholar]
- 57.Olvera A, Cerdà-Cuéllar M, and Aragon V, Study of the population structure of Haemophilus parasuis by multilocus sequence typing. Microbiology (Reading), 2006. 152(Pt 12): p. 3683–3690. [DOI] [PubMed] [Google Scholar]
- 58.Shang P, et al. , Environmental exposure to swine farms reshapes human gut microbiota. Chemosphere, 2022. 307: p. 135558. [DOI] [PubMed] [Google Scholar]
- 59.Møller SA, et al. , Work clothes as a vector for microorganisms: Accumulation, transport, and resuspension of microorganisms as demonstrated for waste collection workers. Environment International, 2022. 161: p. 107112. [DOI] [PubMed] [Google Scholar]
- 60.Barberán A, et al. , Continental-scale distributions of dust-associated bacteria and fungi. Proc Natl Acad Sci U S A, 2015. 112(18): p. 5756–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Shan Y, et al. , Modern urbanization has reshaped the bacterial microbiome profiles of house dust in domestic environments. World Allergy Organ J, 2020. 13(8): p. 100452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Srinivas P, et al. , Beyond genomics in Patescibacteria: A trove of unexplored biology packed into ultrasmall bacteria. Proc Natl Acad Sci U S A, 2024. 121(51): p. e2419369121. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Metagenome data used for this analysis are available at the Sequence Read Archive (SRA) under project number PRJNA975673 (https://www.ncbi.nlm.nih.gov/sra/).