D in the outside frame. doi:10.1371/journal.pone.0053779.gin the reactor sludge metagenome. Relative abundance of the SEED subsystems was shown in Figure S4. Figure S5 demonstrated the comparison of Bacteria and Archaea in the SEED subsystems on Carbohydrate metabolism (Figure S5) and One-carbon metabolism (Figure S5 insert). The number of reads assigned to a specific subsystem (primary y axis) indicates the relative contribution of the domain in the corresponding function category, whereas the percentage of reads assigned 
(secondary y axis) represents the domain’s preference to the functional category. As shown in Figure S5, considering the evident dominance of Bacteria in the community, it is not surprising to find that Bacteria played an important role in all subsystems involved in carbohydrate metabolism except for the one-carbon metabolism in which Archaea was crucial as it exclusively contributed to the methanogenesis process (Figure S5 insert). Additionally, the functions of genera Clostridium and Thermoanaerbacterium were studied in the same manner. Clostridium in the sludge metagenome showed stronger selection in degrading polysaccharides and di-and oligosaccharides while Thermoanaerobacterium preferred more on metabolizing monosaccharides (Figure S6). The KEGG methanogenesis modules were shown in Figure 3; the complete pathway of “Format/Hydrogen/CO2 to methane” and “Methanol to methane” was revealed in the consortia while the acetyl-CoA decarbonylase/synthase complex subunit alpha [EC:1.2.99.2] (shown as box of “Cdh, A,B,C,D,E” in Figure 3) was not observed indicating unfavorable “Acetate to methane”process of the consortia. In addition, the high proportion of formate dehydrogenase [EC:1.2.1.2] and formylmethanofuran dehydrogenase subunit A [EC:1.2.99.5] (Figure 3 insert) pointed out a more active metabolizing of formate/hydrogen/CO2 to methane in the thermophilic sludge consortium. Comparing to the functional annotation by assembled ORFs, many key enzymes in the methane production process especially the enzymes involved in the “Co-enzyme M synthesis module” was missing in the read annotation, indicating that short reads was not suitable for functional analysis of metagenome due to the low annotation efficiency.Mining of Thermo-stable Carbohydrate-active Genes in the Sludge MetagenomeTo identify candidate carbohydrate-active genes from the sludge metagenome, we performed de novo assembly and predicted 31,499 ORFs with an average length of 852 bp and 64 of the 31,499 ORFs were predicted to 94-09-7 chemical information represent full-length genes. To examine the validity of the de novo assembly, we experimentally testified a random subset of 10 putative carbohydrate-active genes (length from 98 to 917 Amino Acids (AA)). The target gene fragments were amplified by specifically designed primers with the DNA extract used to generate the metagenomic data as PCR template. 
Using single set of PCR condition, we obtained 9 out of 10 candidate genes (90 ) with the predicted size (Figure S7). ForMetagenomic Mining of Cellulolytic GenesFigure 2. Taxonomy classification of the metagenome at class level based on RMORF approach. ORFs were assigned by default MEGAN LCA algorithm; only nodes with over 5 ORFs and 1000 reads assigned are shown. The circles are drawn based on the number of reads assigned to the particular node. The number after description denotes, respectively, the sum of reads and ORFs assigned below the particular node. The circles are Eledoisin colored.D in the outside frame. doi:10.1371/journal.pone.0053779.gin the reactor sludge metagenome. Relative abundance of the SEED subsystems was shown in Figure S4. Figure S5 demonstrated the comparison of Bacteria and Archaea in the SEED subsystems on Carbohydrate metabolism (Figure S5) and One-carbon metabolism (Figure S5 insert). The number of reads assigned to a specific subsystem (primary y axis) indicates the relative contribution of the domain in the corresponding function category, whereas the percentage of reads assigned (secondary y axis) represents the domain’s preference to the functional category. As shown in Figure S5, considering the evident dominance of Bacteria in the community, it is not surprising to find that Bacteria played an important role in all subsystems involved in carbohydrate metabolism except for the one-carbon metabolism in which Archaea was crucial as it exclusively contributed to the methanogenesis process (Figure S5 insert). Additionally, the functions of genera Clostridium and Thermoanaerbacterium were studied in the same manner. Clostridium in the sludge metagenome showed stronger selection in degrading polysaccharides and di-and oligosaccharides while Thermoanaerobacterium preferred more on metabolizing monosaccharides (Figure S6). The KEGG methanogenesis modules were shown in Figure 3; the complete pathway of “Format/Hydrogen/CO2 to methane” and “Methanol to methane” was revealed in the consortia while the acetyl-CoA decarbonylase/synthase complex subunit alpha [EC:1.2.99.2] (shown as box of “Cdh, A,B,C,D,E” in Figure 3) was not observed indicating unfavorable “Acetate to methane”process of the consortia. In addition, the high proportion of formate dehydrogenase [EC:1.2.1.2] and formylmethanofuran dehydrogenase subunit A [EC:1.2.99.5] (Figure 3 insert) pointed out a more active metabolizing of formate/hydrogen/CO2 to methane in the thermophilic sludge consortium. Comparing to the functional annotation by assembled ORFs, many key enzymes in the methane production process especially the enzymes involved in the “Co-enzyme M synthesis module” was missing in the read annotation, indicating that short reads was not suitable for functional analysis of metagenome due to the low annotation efficiency.Mining of Thermo-stable Carbohydrate-active Genes in the Sludge MetagenomeTo identify candidate carbohydrate-active genes from the sludge metagenome, we performed de novo assembly and predicted 31,499 ORFs with an average length of 852 bp and 64 of the 31,499 ORFs were predicted to represent full-length genes. To examine the validity of the de novo assembly, we experimentally testified a random subset of 10 putative carbohydrate-active genes (length from 98 to 917 Amino Acids (AA)). The target gene fragments were amplified by specifically designed primers with the DNA extract used to generate the metagenomic data as PCR template. Using single set of PCR condition, we obtained 9 out of 10 candidate genes (90 ) with the predicted size (Figure S7). ForMetagenomic Mining of Cellulolytic GenesFigure 2. Taxonomy classification of the metagenome at class level based on RMORF approach. ORFs were assigned by default MEGAN LCA algorithm; only nodes with over 5 ORFs and 1000 reads assigned are shown. The circles are drawn based on the number of reads assigned to the particular node. The number after description denotes, respectively, the sum of reads and ORFs assigned below the particular node. The circles are colored.