Improved removal performance for NH3 and H2S
Our study evaluated the improved performance in removing odorous gases containing NH3 and H2S. After treatment, the emission concentrations of H2S decreased from 35.3 mg.m− 3 to 0.32 mg.m− 3 in original version and to 0.14 mg.m− 3 in improved version; the emission rate of H2S were decreased from 0.26 kg.h− 1 to 0.0032 kg.h− 1 in original version and to 0.0019 kg.h− 1 in improved version. After treatment, the emission concentrations of NH3 decreased from 16.1 mg.m− 3 to 2.2 mg.m− 3 in original version and to 1.1 mg.m− 3 in improved version; the emission rate of NH3 were decreased from 0.12 kg.h− 1 to 0.022 kg.h− 1 in original version and to 0.009 kg.h− 1 in improved version.
Detail information of the metagenomes
A total of 6.18E + 08 and 5.53E + 08 raw reads were obtained from the CS and TS samples, respectively. Then, 5.90E + 08 clean reads from CS and 5.30E + 08 clean reads from TS were used to assemble metagenomes (Additional file 1). After filtering out the low quality sequences, 143,441 contigs (N50: 1689 bp) were obtained from the CS sample and 155,721 contigs (N50: 1476 bp) were obtained from the TS sample (Fig. 1a and b). Correlation coefficient analysis indicated the sequencing data had good repeatability (Additional file 2). To get an overview of the metagenomic variations, a PCA was performed, and the percentages of explained value in the analysis of PC1 and PC2 were 99.12% and 0.44%, respectively (Fig. 1c).
Functional prediction and classification of unigenes
Based on the assembled contigs, a total of 901,016 unigenes with an average length of 660 bp and GC content of 58% were predicted. All the unigenes were detected in the CS sample and only 875,737 unigenes were detected in the TS sample (Fig. 1d). Length distribution of all predicted unigenes was showed in Additional file 3 and expression abundances of unigenes were showed in Additional file 4.
In total, 726,022 unigenes were assigned to different functional KEGG pathways. In the ‘metabolism’ category, the most typical pathways were ‘carbohydrate metabolism’ (59,872 unigenes), ‘amino acid metabolism’ (49,584 unigenes), and ‘cofactors and vitamins metabolism’ (30,315 unigenes). In the ‘genetic information processing’ category, most unigenes were classed into the ‘translation’ (18,585 unigenes), ‘replication and repair’ (15,850 unigenes), and ‘folding, sorting and degradation’ (11,594 unigenes) pathways. In the ‘environmental information processing’ category, most unigenes were grouped into the ‘signal transduction’ (27,557 unigenes) and ‘membrane transport’ (26,101 unigenes). In the ‘cellular processes’ category, the major terms were ‘cell motility’ (19,559 unigenes) and ‘cellular community’ (9,011 unigenes) (Additional file 4).
Taxonomic profile of the two metagenomes
Based on the predicted ORFs, the taxonomy annotation and abundance of microbial species derived from the two sample groups was analyzed. At phylum level, 184 taxa were summarized from the two sample groups (Additional file 5) According to their annotation, Proteobacteria (37.7%), Planctomycetes (8.7%), Chloroflexi (2.5%), Bacteroidetes (2.2%), Cyanobacteria (1.6%), and Actinobacteria (1.0%) were considered to be the dominant phyla, accounting for more than 1% of the total population. At the phyla level, the significantly up-regulated taxa are Methanosarcinaceae (3.55 fold) and Ichthyobacteriaceae (4.78 fold), and the significantly down-regulated taxa are Morchellaceae (-2.26 fold), Nautiliaceae (-2.28 fold) and Sterolibacteriaceae (-1.75 fold).
At the genus level, a total of 3619 genera were obtained from the two sample groups. Among these genera, there are 112 genera with a relative abundance more than 0.1% of the total microbes. Microbial compositions for both CS and TS sample groups at genus level were showed in Fig. 2a. According to their annotation, Thioalkalivibrio (2.73%), Thauera (1.94%), and Pseudomonas (0.1%) were considered to be the dominant genera, accounting for more than 0.1% of the total population (Additional file 6). Several dominant genera, including Methanomethylovorans (3.91 fold), Stappia (0.83 fold), Mesorhizobium (0,2 fold), Desulfovibrio (0.32 fold), Mesotoga (0.87 fold), and Defluviicoccus (0.31 fold), were significantly up-regulated during the treatment process. Contrarily, several other dominant genera, such as Methyloversatilis (-1.82 fold), Elioraea (-0.66 fold), Rhodovulum (-0.82 fold), and Amaricoccus (-0.92 fold), were significantly down-regulated during the treatment process (Fig. 2b).
Analysis of the DEGs between CS and TS samples
A large number of DEGs, including 166,011 up-regulated and 151,567 down-regulated genes, were identified in our study (Fig. 3a). According to their annotations, most of the DEGs were assigned into different categories. For the GO classification, the top five significant enriched GO terms were ‘symporter activity’, ‘phosphorelay signal transduction system’, ‘serine-type carboxypeptidase activity’, ‘peptide metabolic process’, and ‘metallocarboxypeptidase activity’ (Fig. 3b). Most of the DEGs were grouped into 187 KEGG metabolic pathways (Additional file 7). Based on their KEGG classification, the top five significant enriched KEGG terms were ‘Two-component system’, ‘Other glycan degradation’, ‘Sphingolipid metabolism’, ‘Galactose metabolism’, and ‘Bacterial chemotaxis’ (Fig. 3c).
Analysis of xenobiotic biodegradation pathway-related KEGGs
Previous studies have reported several xenobiotic biodegradation pathways in different microbes . In our study, a large number of DEGs involved in 15 typical xenobiotic biodegradation pathways were identified (Additional file 8). The benzoate degradation pathway (map00363) contained the largest number of DEGs, including 310 up- and 550 down-regulated genes. The second largest xenobiotic biodegradation pathway was the chloroalkane and chloroalkene degradation pathway, including 252 up- and 228 down-regulated genes. Aminobenzoate degradation pathway was the third largest xenobiotic biodegradation pathway, containing 125 up- and 198 down-regulated genes (Fig. 4).
Analysis of the genes involved in the nitrogen metabolic and sulfur metabolic pathways
A number of genes involved in the nitrogen metabolic and sulfur metabolic pathways have been reported in the past years . For the nitrogen metabolism, the genes encoding five key enzymes, including nirK, nirB, nrfA, hao, and nirA, were identified in our study (Fig. 5a). For the sulfur metabolism, the genes encoding seven key enzymes, including suoX, dsrA, sir1, asr1, glpE, phsA, and fccB, were identified in our study (Fig. 5b). For the nitrogen metabolic pathway, most of the nirA and hao encoding genes were significantly up-regulated during the treatment (Fig. 5c). For the sulfur metabolic pathway, most of the phsA encoding genes were up-regulated and most of the suoX encoding genes were down-regulated during the treatment (Fig. 5d).
Comparison of the microbial community between original and improved biofilters
In our study, comparison of the microbial community between original and improved biofilters has been performed. Firstly, we analyzed the changes between original and improved biofilters at phyla level. In the original biofilter, the significantly up-regulated phyla were Proteobacteria, Euryarchaeota and Nitrospirae, and the significantly down-regulated phyla were Ignavibacteriae, Bacteroidetes, and Planctomycetes (Fig. 6a). In the improved biofilter, the significantly up-regulated phyla were Deferribacteres, Tenericutes, and Microsporidia, which showed opposite responses in the original version. While in the improved biofilter, the significantly down-regulated phyla were Elusimicrobia, Fibrobacteres, and Verrucomicrobia, which showed similar responses in the original version (Fig. 6b).
Then, we analyzed the changes between original and improved biofilters at genera level. In the original biofilter, the significantly changed genera were Ferroplasma and Cetobacterium, which showed no responses in the improved version (Fig. 6c). In the improved biofilter, the most up-regulated genera was Arcanobacterium, which was siginificantly down-regulated in the original version, and the most down-regulated genera was Oleispira, which showed similar response in the original version (Fig. 6d).
Comparison of the functional genes between original and improved biofilters
Comparison of the functional genes between original and improved biofilters has been also performed. For the nitrogen metabolic pathway, average expression levels of the nirA, nifA and nirB encoding genes were significantly up-regulated in the improved biofilter and only the nirB encoding genes were up-regulated in the original version (Fig. 7a). For the sulfur metabolic pathway, most of the key genes were up-regulated in both of two biofilters, except for phsA and fccB. The average expression levels of the phsA and fccB encoding genes were significantly increased in the improved version (Fig. 7b).