Computational protocol: Restructuring of Epibacterial Communities on Fucus vesiculosus forma mytili in Response to Elevated pCO2 and Increased Temperature Levels

Similar protocols

Protocol publication

[…] Sequence processing was performed using mothur version 1.34.4 (Schloss et al., ; Kozich et al., ). Raw reads were concatenated to 9,364,598 contiguous sequences (contigs) using the command make.contig. Contigs with ambiguous bases or homopolymers longer than 8 bases as well as contigs longer than 552 bases were removed using screen.seqs. The remaining 8,354,564 contigs were screened for redundant sequences using unique.seqs and clustered into 2,537,511 unique sequences. The sequences were consecutively aligned (with align.seqs) to a modified version of the SILVA database release 102 (Pruesse et al., ) containing only the hypervariable regions V1 and V2. Sequences not aligning in the expected region were removed from the dataset with screen.seqs. The alignment was condensed by removing gap-only columns with filter.seqs. The final alignment contained 8,288,816 sequences (2,511,577 unique) of lengths between 253 and 450 bases. Rare and closely related sequences were clustered using unique.seqs and precluster.seqs. The latter was used to include sequences with up to three positional differences compared to larger sequence clusters into the latter. Chimeric sequences were removed using the Uchime algorithm (Edgar et al., ) via the command chimera.uchime, followed by remove.seqs. This left 7,934,922 sequences (163,446 unique) in the dataset. Sequence classification was performed using the Wang Method (Wang et al., ) on a modified Greengenes database (containing only the hypervariable regions V1 and V2) with a bootstrap threshold of 80%. Sequences belonging to the kingdom archaea, to chloroplasts or mitochondria were removed from the dataset using remove.lineage. OTUs (operational taxonomic units) were formed using the average neighbor clustering method with cluster.split. Parallelization of this step was done taking the taxonomic classification on the order level into account. A sample-by-OTU table containing 55,378 OTUs at the 97% level was generated using make.shared. OTUs were classified taxonomically using the modified Greengenes database mentioned above and the command classify.otu. [...] Statistical downstream analysis was performed with custom scripts in R v3.1.3 (R Core Team, ). OTUs of very low abundance only increase computation time without contributing useful information. They were thus removed from the dataset as follows: After transformation of counts in the sample-by-OTU table to relative abundances (based on the total number of reads per sample), OTUs were ordered by decreasing mean percentage across samples. The set of ordered OTUs for which the cumulative mean percentage amounted to 95% was retained in the filtered OTU table, resulting in a decrease in the number of OTUs from 55,378 to 4,157.The extent of change in relative OTU abundance across samples explained by the experimental factors Temperature, pCO2, Time, and Sample Type (see Table ) was explored by redundancy analysis (RDA) with Hellinger-transformed OTU counts (Stratil et al., , ; Langfeldt et al., ) using function rda of R package vegan v2.4-0 (Oksanen et al., ).Model selection started with a full RDA model containing all main effects and interactions of experimental factors, using the following model formula: Transformed OTU counts~(Type·Temp·CO2)                                                             + (Type·Temp·CO2):WeekThe Week main effect was omitted from the formula as temporal effects were nested within other levels or their combinations. (Inclusion of a Week main effect would be justified if measures for factor Week were completely independent rather than repeated). A permutation scheme for permutation-based significance tests was chosen with function how of R package permute v0.8-4 (Simpson, ) to reflect the repeated-measures design as well as the temporal nature of factor Week. Permutation of samples within a sample unit (a set of repeated measures taken for a particular combination of Type, Temperature and pCO2 at the different time points of Week; see Table ) was set to “series,” with the same permutation used for each sample unit; clusters of samples belonging to different sample units were allowed to be permuted freely.The full model was simplified by backward selection with function ordistep. The final RDA model exhibited significant interaction effects CO2:Week and Type:Temp:Week (see Results Section). It was thus necessary to evaluate (i) the Temperature effect within each level of Type and Week and (ii) the pCO2 treatment effect within each level of Week. For (i) the effect of the Temperature:Week interaction was evaluated within each level of factor Type in appropriate submodels of the final RDA model; upon significance, the Temperature effect was further evaluated within each level of factor Week (within the same level of factor Type). For (ii) the pCO2 effect was evaluated within each level of factor Week in appropriate submodels of the final RDA model. p-values for each stage of these hierarchical testing schemes were corrected for multiple testing by Benjamini–Hochberg correction (false discovery rate, FDR; Benjamini and Hochberg, ). For each significant submodel, OTUs were determined that were significantly correlated with any axis in the RDA submodel by function envfit with 105 permutations, followed by Benjamini–Hochberg correction. In order to reduce the number of tests in this procedure, OTUs were pre-filtered according to their vector lengths calculated from corresponding RDA scores (scaling 1) by profile likelihood selection (Zhu and Ghodsi, ).OTUs significant at an FDR of 5% were further subject to indicator analysis with function multipatt of the R package indicspecies v1.7.5 (De Cáceres and Legendre, ) with 105 permutations. Indicator OTUs (iOTUs)—in analogy to indicator species sensu De Cáceres and Legendre ()—are OTUs that prevail in a certain sample group (here: either a level of pCO2 within a certain level of Week, or a level of Temperature within a certain sample Type and level of Week) while being found only irregularly and at low abundance in other sample groups.3D visualizations of the final RDA model were produced in kinemage format (Richardson and Richardson, ) using the R package R2Kinemage developed by S.C.N., and displayed in KiNG v2.21 (Chen et al., ). For alpha diversity analysis, effective OTU richness (Shannon numbers equivalent, 1D; Jost, , ) was calculated from the filtered OTU table. Multi-panel alpha diversity and iOTU plots were drawn with R package lattice v0.20-33 (Sarkar, ).An OTU association network was inferred with the R package SpiecEasi v0.1 (Kurtz et al., ). OTUs selected for network analysis were required to be present in ≥60% of all samples to reduce the number of zeros in the data for a robust calculation, resulting in a set of 97 OTUs. Calculations were performed with the Meinshausen and Bühlmann neighborhood selection framework (MB method; Meinshausen and Bühlmann, ). Correlations between associated OTUs were determined from the centered log-ratio-transformed counts. The igraph package v1.0.1 (Csárdi and Nepusz, ) was employed for visualizing moderate to strong OTU associations (absolute correlation ≥0.6). [...] In order to evaluate the interaction effect of temperature and pCO2 on all variables measured (relative biomass growth rates, C:N ratio, C, N and mannitol content) at the end of the experiment, two-way ANOVAs were used with temperature and pCO2 as fixed factors. When the analysis did not show significant interactions, a one-way ANOVA was carried out for each factor separately. When the one-way ANOVA revealed significant differences, a post hoc Tukey's honest significant difference test was applied. Prior to the use of ANOVAs, data were tested for normality with the Kolmogorov-Smirnov and for homogeneity with the Levene's test. Data were analyzed using SPSS Statistics 20 (IBM, Armonk, NY, USA). […]

Pipeline specifications

Software tools mothur, UCHIME, Igraph, SPSS
Applications Miscellaneous, 16S rRNA-seq analysis
Organisms Fucus vesiculosus
Chemicals Carbon Dioxide