--- title: "Schuck_animal_microbiome" author: "Laura Schuck" date: "`r Sys.Date()`" output: html_document --- ```{r setup, include=FALSE} knitr::opts_chunk$set(echo = TRUE) ``` ###Bioinformatics We received filtered data from the sequencing facility. The filtering thresholds are described in the methods section. Amplicon sequence data is available under NCBI (PRJNA1354894), isolates sequences are available in the AmphiBac database (https://github.com/AmphiBac). All data is also openly available at request to the correspondent author (laurakauerschuck@gmail.com) ```{bash bionformatics} module load anaconda conda activate qiime2-amplicon-2023.9 cd /storage/work/lxk5406/masters2 #gzip -d Rwads/*.gz qiime tools import --type 'SampleData[SequencesWithQuality]' --input-path sample_list.tsv --output-path samples.qza --input-format SingleEndFastqManifestPhred33V2 qiime dada2 denoise-single \ --i-demultiplexed-seqs samples.qza \ --p-trunc-len 0 \ --p-n-threads 8 \ --p-n-reads-learn 50000 \ --output-dir dada2_output #visualize dada2 table qiime metadata tabulate \ --m-input-file dada2_output/denoising_stats.qza \ --o-visualization stats-dada2.qzv qiime tools export --input-path dada2_output/denoising_stats.qza --output-path dada2_output qiime tools export --input-path dada2_output/table.qza --output-path dada2_output_exported qiime tools export --input-path dada2_output/representative_sequences.qza --output-path dada2_output_exported biom convert -i dada2_output_exported/feature-table.biom -o dada2_output_exported/feature-table.txt --to-tsv #change file names to match later script mv /storage/work/lxk5406/masters2/dada2_output/representative_sequences.qza rep-seqs.qza mv /storage/work/lxk5406/masters2/dada2_output/table.qza table.qza #summarize the feature table qiime feature-table summarize \ --i-table table.qza \ --m-sample-metadata-file metadata2.tsv \ --o-visualization table.qzv qiime feature-table tabulate-seqs \ --i-data rep-seqs.qza \ --o-visualization rep-seqs.qzv #align sequences qiime alignment mafft \ --i-sequences ./rep-seqs.qza \ --o-alignment seq_dada2_aligned.qza #filter alignment qiime alignment mask \ --i-alignment ./seq_dada2_aligned.qza \ --o-masked-alignment seq_dada2_aligned_filtered.qza #create phylogenetic tree qiime phylogeny fasttree \ --i-alignment ./seq_dada2_aligned_filtered.qza \ --o-tree tree_unrooted.qza #root the phylogenetic tree qiime phylogeny midpoint-root \ --i-tree ./tree_unrooted.qza \ --o-rooted-tree tree_rooted.qza ##tree completed #download trained Bayesian classifier #wget -O "gg_2022_10_backbone_full_length.nb.qza" https://data.qiime2.org/classifiers/greengenes/gg_2022_10_backbone_full_length.nb.qza #assign taxonomy using classifier qiime feature-classifier classify-sklearn \ --i-classifier gg_2022_10_backbone_full_length.nb.qza \ --i-reads ./rep-seqs.qza \ --o-classification dada2_taxonomy.qza #export taxonomy to get .tsv file with taxonomy for each feature qiime tools export \ --input-path ./dada2_taxonomy.qza \ --output-path dada2_taxonomy #filter OTU table based on taxonomy to remove mitochondria and chloroplasts qiime taxa filter-table \ --i-table ./table.qza \ --i-taxonomy ./dada2_taxonomy.qza \ --p-exclude mitochondria,chloroplast,Unassigned \ --p-mode contains \ --o-filtered-table table_filt2.qza #filtered OTU table qiime feature-table summarize \ --i-table ./table_filt2.qza \ --m-sample-metadata-file metadata2.tsv \ --o-visualization table_filt2.qzv #view rarefaction curves qiime diversity alpha-rarefaction \ --i-table ./table_filt2.qza \ --p-max-depth 5620 \ --p-metrics observed_features \ --o-visualization alpha_rarefaction_r5620.qzv \ --m-metadata-file ./metadata2.tsv #rarefy to 5620 qiime feature-table rarefy \ --i-table ./table_filt2.qza \ --p-sampling-depth 5620 \ --o-rarefied-table table_filt_r5620.qza qiime feature-table summarize \ --i-table ./table_filt_r5620.qza \ --m-sample-metadata-file ./metadata2.tsv \ --o-visualization table_filt_r5620.qzv #get taxa barplots qiime taxa barplot \ --i-table ./table_filt_r5620.qza \ --i-taxonomy ./dada2_taxonomy.qza \ --m-metadata-file ./metadata2.tsv \ --output-dir taxonomy_plots_r5620 #alpha diversity qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620 qiime tools export \ --input-path ./core_metrics_r5620/observed_features_vector.qza \ --output-path observed_features_r5620 qiime tools export \ --input-path ./core_metrics_r5620/faith_pd_vector.qza \ --output-path faith_pd_r5620 qiime tools export \ --input-path ./core_metrics_r5620/evenness_vector.qza \ --output-path evenness_r5620 qiime tools export \ --input-path ./core_metrics_r5620/shannon_vector.qza \ --output-path shannon_r5620 #Beta Diversity qiime tools export \ --input-path ./core_metrics_r5620/bray_curtis_distance_matrix.qza \ --output-path bray_curtis_dm_r5620 qiime tools export \ --input-path ./core_metrics_r5620/bray_curtis_pcoa_results.qza \ --output-path bray_curtis_pcoa_r5620 qiime tools export \ --input-path ./core_metrics_r5620/jaccard_distance_matrix.qza \ --output-path jaccard_dm_r5620 qiime tools export \ --input-path ./core_metrics_r5620/jaccard_pcoa_results.qza \ --output-path jaccard_pcoa_r5620 qiime tools export \ --input-path ./core_metrics_r5620/unweighted_unifrac_distance_matrix.qza \ --output-path unweighted_unifrac_dm_r5620 qiime tools export \ --input-path ./core_metrics_r5620/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620 qiime tools export \ --input-path ./core_metrics_r5620/weighted_unifrac_distance_matrix.qza \ --output-path weighted_unifrac_dm_r5620 qiime tools export \ --input-path ./core_metrics_r5620/weighted_unifrac_pcoa_results.qza \ --output-path weighted_unifrac_pcoa_r5620 #filter by host species qiime feature-table filter-samples \ --i-table ./table_filt_r5620.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "condition='Haddadus binotatus'" \ --o-filtered-table table_filt_r5620_had.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "condition='Ischnocnema henselii'" \ --o-filtered-table table_filt_r5620_hen.qza \ --p-no-exclude-ids #calculate core microbiome qiime feature-table core-features \ --i-table ./table_filt_r5620_had.qza \ --p-min-fraction 0.90 \ --output-dir ./core_microbiome_r5620_had qiime feature-table core-features \ --i-table ./table_filt_r5620_hen.qza \ --p-min-fraction 0.90 \ --output-dir ./core_microbiome_r5620_hen #NOTHING FOUND #core diversity qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_had.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_had qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_hen.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_hen #Beta Diversity Haddadus qiime tools export \ --input-path ./core_metrics_r5620_had/bray_curtis_distance_matrix.qza \ --output-path bray_curtis_dm_r5620_had qiime tools export \ --input-path ./core_metrics_r5620_had/bray_curtis_pcoa_results.qza \ --output-path bray_curtis_pcoa_r5620_had qiime tools export \ --input-path ./core_metrics_r5620_had/jaccard_distance_matrix.qza \ --output-path jaccard_dm_r5620_had qiime tools export \ --input-path ./core_metrics_r5620_had/jaccard_pcoa_results.qza \ --output-path jaccard_pcoa_r5620_had qiime tools export \ --input-path ./core_metrics_r5620_had/unweighted_unifrac_distance_matrix.qza \ --output-path unweighted_unifrac_dm_r5620_had qiime tools export \ --input-path ./core_metrics_r5620_had/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_had qiime tools export \ --input-path ./core_metrics_r5620_had/weighted_unifrac_distance_matrix.qza \ --output-path weighted_unifrac_dm_r5620_had --input-path ./core_metrics_r5620_had/weighted_unifrac_pcoa_results.qza \ --output-path weighted_unifrac_pcoa_r5620_had #Beta Diversity henselii qiime tools export \ --input-path ./core_metrics_r5620_hen/bray_curtis_distance_matrix.qza \ --output-path bray_curtis_dm_r5620_hen qiime tools export \ --input-path ./core_metrics_r5620_hen/bray_curtis_pcoa_results.qza \ --output-path bray_curtis_pcoa_r5620_hen qiime tools export \ --input-path ./core_metrics_r5620_hen/jaccard_distance_matrix.qza \ --output-path jaccard_dm_r5620_hen qiime tools export \ --input-path ./core_metrics_r5620_hen/jaccard_pcoa_results.qza \ --output-path jaccard_pcoa_r5620_hen qiime tools export \ --input-path ./core_metrics_r5620_hen/unweighted_unifrac_distance_matrix.qza \ --output-path unweighted_unifrac_dm_r5620_hen qiime tools export \ --input-path ./core_metrics_r5620_hen/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_hen qiime tools export \ --input-path ./core_metrics_r5620_hen/weighted_unifrac_distance_matrix.qza \ --output-path weighted_unifrac_dm_r5620_hen qiime tools export \ --input-path ./core_metrics_r5620_hen/weighted_unifrac_pcoa_results.qza \ --output-path weighted_unifrac_pcoa_r5620_hen #filter by host species #qiime feature-table filter-samples \ # --i-table ./table_filt_r5620.qza \ # --m-metadata-file ./metadata2.tsv \ # --p-where "condition='Haddadus binotatus'" \ # --o-filtered-table table_filt_r5620_had.qza \ # --p-no-exclude-ids #filter by host species and time qiime feature-table filter-samples \ --i-table ./table_filt_r5620_had.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='field'" \ --o-filtered-table table_filt_r5620_had_field.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620_had.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='post_ant'" \ --o-filtered-table table_filt_r5620_had_pant.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620_had.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='post_bd'" \ --o-filtered-table table_filt_r5620_had_pbd.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620_had.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='final'" \ --o-filtered-table table_filt_r5620_had_final.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620_had.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='post_mortem'" \ --o-filtered-table table_filt_r5620_had_pmortem.qza \ --p-no-exclude-ids qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_had_field.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_had_field qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_had_pant.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_had_pant qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_had_pbd.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_had_pbd qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_had_final.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_had_final qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_had_pmortem.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_had_pmortem qiime tools export \ --input-path ./core_metrics_r5620_had_field/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_had_field qiime tools export \ --input-path ./core_metrics_r5620_had_pant/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_had_pant qiime tools export \ --input-path ./core_metrics_r5620_had_pbd/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_had_pbd qiime tools export \ --input-path ./core_metrics_r5620_had_final/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_had_final qiime tools export \ --input-path ./core_metrics_r5620_had_pmortem/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_had_pmortem qiime feature-table filter-samples \ --i-table ./table_filt_r5620.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "condition='Ischnocnema henselii'" \ --o-filtered-table table_filt_r5620_hen.qza \ --p-no-exclude-ids #filter by host species and time qiime feature-table filter-samples \ --i-table ./table_filt_r5620_hen.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='field'" \ --o-filtered-table table_filt_r5620_hen_field.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620_hen.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='post_ant'" \ --o-filtered-table table_filt_r5620_hen_pant.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620_hen.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='post_bd'" \ --o-filtered-table table_filt_r5620_hen_pbd.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620_hen.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='final'" \ --o-filtered-table table_filt_r5620_hen_final.qza \ --p-no-exclude-ids qiime feature-table filter-samples \ --i-table ./table_filt_r5620_hen.qza \ --m-metadata-file ./metadata2.tsv \ --p-where "time='post_mortem'" \ --o-filtered-table table_filt_r5620_hen_pmortem.qza \ --p-no-exclude-ids qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_hen_field.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_hen_field qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_hen_pant.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_hen_pant qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_hen_pbd.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_hen_pbd qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_hen_final.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_hen_final qiime diversity core-metrics-phylogenetic \ --i-phylogeny ./tree_rooted.qza \ --i-table ./table_filt_r5620_hen_pmortem.qza \ --p-sampling-depth 5620 \ --m-metadata-file ./metadata2.tsv \ --output-dir core_metrics_r5620_hen_pmortem qiime tools export \ --input-path ./core_metrics_r5620_hen_field/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_hen_field qiime tools export \ --input-path ./core_metrics_r5620_hen_pant/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_hen_pant qiime tools export \ --input-path ./core_metrics_r5620_hen_pbd/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_hen_pbd qiime tools export \ --input-path ./core_metrics_r5620_hen_final/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_hen_final qiime tools export \ --input-path ./core_metrics_r5620_hen_pmortem/unweighted_unifrac_pcoa_results.qza \ --output-path unweighted_unifrac_pcoa_r5620_hen_pmortem ``` I merged the diversity results with the metadata. Decidaded to use unweighted unifrac for more representativity of rare OTUs and to consider phylogeny. ##Isolate sequences First I used this script to define which isolates I would consider Bd-inhibitory" ```{r inhibition} setwd("/Users/lauraschuck/Desktop/Bioassays") data <- read.csv("testeR.csv", header = TRUE) data$Plate <- as.factor(data$Plate) data$Day <- as.factor(data$Day) library(dplyr) # Calculate the mean absorbance for each triplicate of a given Sample, Plate, and Day mean_data <- data %>% group_by(Plate, Day, Sample) %>% summarize(across(starts_with("Absorbance"), mean, na.rm = TRUE)) %>% ungroup() # Write the result to a new CSV file write.csv(mean_data, "mean_absorbance_data.csv", row.names = FALSE) # Define the output directory and file path output_dir <- "/Users/lauraschuck/Desktop/Bioassays/mnt/data" # Modify this path as needed output_file <- file.path(output_dir, "mean_absorbance_data.csv") # Load necessary library library(dplyr) # Read the CSV file data <- mean_data # Calculate the mean absorbance of the negative control (Heat killed Bd) for each plate HK_mean <- data %>% filter(Sample == "HK-Bd0001") %>% group_by(Plate) %>% summarize(mean_absorbance = mean(Absorbance)) # Correct the absorbance values by subtracting the mean HK absorbance for each plate data_corrected <- data %>% left_join(HK_mean, by = "Plate") %>% mutate(Corrected_Absorbance = Absorbance - mean_absorbance) %>% select(-mean_absorbance) # Separate data into day "zero" and "final" day_zero <- data_corrected %>% filter(Day == "zero") %>% select(Plate, Sample, Corrected_Absorbance) %>% rename(Absorbance_zero = Corrected_Absorbance) day_final <- data_corrected %>% filter(Day == "final") %>% select(Plate, Sample, Corrected_Absorbance) %>% rename(Absorbance_final = Corrected_Absorbance) # Merge the data to align day "zero" and "final" samples merged_data <- merge(day_final, day_zero, by = c("Plate", "Sample")) # Calculate the corrected absorbance for the final day by subtracting the zero day absorbance corrected_data <- merged_data %>% mutate(Corrected_Absorbance = Absorbance_final - Absorbance_zero) # Calculate the corrected Positive0001 value for each plate positive_values <- corrected_data %>% filter(Sample == "Positive0001") %>% select(Plate, Corrected_Absorbance) %>% rename(Positive_Corrected_Absorbance = Corrected_Absorbance) # Merge the positive values with the corrected data corrected_data <- corrected_data %>% left_join(positive_values, by = "Plate") # Calculate the Proportion and Inhibition corrected_data <- corrected_data %>% mutate( Proportion = Corrected_Absorbance / Positive_Corrected_Absorbance, Inhibition = 1 - Proportion ) # Create the result column corrected_data <- corrected_data %>% mutate( result = ifelse(Inhibition > 0.50, "Inhibitor", "Non-inhibitor") ) # Exclude Positive0001 and select relevant columns, then order by Sample final_data <- corrected_data %>% filter(Sample != "Positive0001") %>% select(Plate, Sample, Inhibition, result) %>% arrange(Sample) # Save the final table write.csv(final_data, "inhibition_results.csv", row.names = FALSE) # Display the final data print(final_data) #Merge with metadata # Load necessary libraries library(dplyr) # Load the datasets isolates_metadata <- read.csv("/Users/lauraschuck/Desktop/Isolates/isolates_metadata.csv") inhibition_results <- read.csv("/Users/lauraschuck/Desktop/Bioassays/inhibition_results.csv") # Merge the datasets by "Sample" and "SampleID" columns merged_data <- merge(isolates_metadata, inhibition_results, by.x = "SampleID", by.y = "Sample") # Save the merged data to a new CSV file write.csv(merged_data, "/Users/lauraschuck/Desktop/Isolates/merged_data.csv", row.names = FALSE) ``` I've used the IsolateR package to filter the isolate sequences and create the inhibitory-only and the full libraries: ```{r isolate_filtering} library(isolateR) folder_list <- c("/path/to/folder_containing_ab1_files1", "/path/to/folder_containing_ab1_files2", "/path/to/folder_containing_ab1_files3") isoALL.S4 <- isoALL(input=folder_list, merge=TRUE) ``` This provided me with Fasta files for each library. Here I describe the matching process: ##Isolate matching #VSearch modification We used VSearch to match our amplicon and isolate libraries. VSEARCH is a sequence search tool used in bioinformatics to compare DNA sequences quickly and efficiently. By default, vsearch limits the number of returned matches for each query (--maxaccepts, default = 1) and stops searching once a set number of non-matches are encountered (-maxrejects, default = 32). These defaults boost speed but may miss valid hits, especially in exhaustive search protocols. Parameter Changes We changed the following: • —maxaccepts 0 Setting this to zero removes the limit on the number of acceptable hits per query. This ensures that all database sequences matching the query at or above the identity threshold (e.g. 100% match) are reported. Use this for comprehensive matching. • -maxrejects 0 Setting this to zero disables the early cutoff of rejected sequences. VSEARCH will continue searching the full database for each query, regardless of how many candidates are rejected. This prevents missing matches after a string of non-matches. Reason for Changes These settings are recommended for exhaustive, conservative analysis—such as matching short 16S V4 amplicon reads to longer bacterial reference sequences with strict identity requirements. • All true matches are reported. • No matches are missed due to default shortcut limits. • More computational resources may be required due to full database searches for every query. Example code (changed within the _cluster_features.py file in q2-vsearch folder): def cluster_features_de_novo(sequences: DNAFASTAFormat, table: biom.Table, perc_identity: float, threads: int = 1 ) -> (biom.Table, DNAFASTAFormat): clustered_sequences = DNAFASTAFormat() with tempfile.NamedTemporaryFile() as fasta_with_sizes: with tempfile.NamedTemporaryFile() as out_uc: _fasta_with_sizes(str(sequences), fasta_with_sizes.name, table) cmd = ['vsearch', '--cluster_size', fasta_with_sizes.name, '--id', str(perc_identity), '--centroids', str(clustered_sequences), '--uc', out_uc.name, '--qmask', 'none', # ensures no lowercase DNA chars '--xsize', '--threads', str(threads), '--minseqlength', '1', '--maxrejects', '0', #changed '--maxaccepts', '0', #changed '--fasta_width', '0'] run_command(cmd) out_uc.seek(0) conn = _uc_to_sqlite(out_uc) collapse_f = _collapse_f_from_sqlite(conn) table = table.collapse(collapse_f, norm=False, min_group_size=1, axis='observation', include_collapsed_metadata=False) return table, clustered_sequences #Running VSearch When done changing, I ran this code using my strict_inhibitory and full libraries: ``` {bash vsearch} module load anaconda conda activate qiime2-amplicon-2023.9 cd /storage/work/lxk5406/matching #ANTI-BD DATABASE MATCHING #table with just rare ASVs filtered #filter sequences to just retained ASVs qiime feature-table filter-seqs \ --i-table table_filt_r5620.qza \ --i-data rep-seqs.qza \ --p-no-exclude-ids \ --o-filtered-data OTUtable-filtered_Repset.qza #match to STRICT INHIBITORY database at 100% similarity qiime vsearch cluster-features-closed-reference \ --i-sequences OTUtable-filtered_Repset.qza \ --i-table table_filt_r5620.qza \ --i-reference-sequences inhibitor_only_lks.qza \ --p-perc-identity 1 \ --p-strand both \ --output-dir Antifungal_Matches_100 qiime feature-table presence-absence \ --i-table Antifungal_Matches_100/clustered_table.qza \ --o-presence-absence-table PresAbs_clustered_table_lks.qza #match to FULL database at 100% similarity qiime vsearch cluster-features-closed-reference \ --i-sequences OTUtable-filtered_Repset.qza \ --i-table table_filt_r5620.qza \ --i-reference-sequences full_db_lks.qza \ --p-perc-identity 1 \ --p-strand both \ --output-dir Cultured_Matches_100 qiime tools export \ --input-path Antifungal_Matches_100/clustered_table.qza \ --output-path clustered_table_exported ``` #Calculating proportion of Bd-inhibitory reads: ```{R matching_proportions} setwd("/Users/lauraschuck/Desktop/submissions/antifungal_predictions/") if (!require("readr")) { install.packages("readr", repos="http://cran.rstudio.com/") library("readr") } if (!require("magrittr")) { install.packages("magrittr", repos="http://cran.rstudio.com/") library("magrittr") } if (!require("dplyr")) { install.packages("dplyr", repos="http://cran.rstudio.com/") library("dplyr") } if (!require("purrr")) { install.packages("purrr", repos="http://cran.rstudio.com/") library("purrr") } if (!require("stringr")) { install.packages("stringr", repos="http://cran.rstudio.com/") library("stringr") } if (!require("devtools")) { install.packages("devtools", repos="http://cran.rstudio.com/") library("devtools") } if (!require("qiime2R")) { install_github("jbisanz/qiime2R") library("qiime2R") } library(tidyr) Antifungal_table = qiime2R::read_qza("Antifungal_Matches_100/clustered_table.qza") #import the clustered table from above (contains the number of reads matching the inhibitory database for each sample before rarefaction) #include full pathname Antifungal_table = qiime2R::read_qza("Antifungal_Matches_100/clustered_table.qza") Antifungal_table = as.data.frame(Antifungal_table$data) TotalAntiFungal = Antifungal_table %>% select_if(is.numeric) %>% map_dbl(sum) TotalAntiFungal = as.data.frame(TotalAntiFungal) TotalAntiFungal = cbind(id = rownames(TotalAntiFungal), TotalAntiFungal) rownames(TotalAntiFungal) <- NULL #import the clustered table from above (contains the number of reads matching the full database for each sample before rarefaction) Cultured_table = qiime2R::read_qza("Cultured_Matches_100/clustered_table.qza") Cultured_table = as.data.frame(Cultured_table$data) TotalCultured = Cultured_table %>% select_if(is.numeric) %>% map_dbl(sum) TotalCultured = as.data.frame(TotalCultured) TotalCultured = cbind(id = rownames(TotalCultured), TotalCultured) rownames(TotalCultured) <- NULL #join the results into one dataframe Antifungal_Cultured_Reads <- left_join(TotalCultured, TotalAntiFungal, by = "id") Antifungal_Cultured_Reads3 <- Antifungal_Cultured_Reads %>% mutate(RatioAF_Full_PR = TotalAntiFungal/TotalCultured) %>% #calculate the ratio of inhibitory/all matching reads mutate(Representation = TotalCultured/5620) %>% #calculate proportion of total reads matching full DB rename(TotalAntiFungal_PR = TotalAntiFungal) %>% #rename the columns so that they are labeled as PR (pre-rarefaction) rename(TotalCultured_PR = TotalCultured) %>% replace_na(list(RatioAF_Full_PR = 0)) #replace NAs with 0s Antifungal_Cultured_Reads3$id <- gsub("-", "\\.", Antifungal_Cultured_Reads3$id) #this just fixes the IDs so they match my metadata write.csv(Antifungal_Cultured_Reads3, "Antifungal_Cultured_Reads3.csv", row.names = F) #write this dataframe so you can merge it with your dataframe Metadata = read.csv("metadata_final.csv") #Metadata$id <- gsub("\\.", "-", Metadata$id) Antifungal_tablePA = qiime2R::read_qza("Antifungal_Matches_100/PresAbs_clustered_table.qza") Antifungal_tablePA = as.data.frame(Antifungal_tablePA$data) AntiFungal_Richness = Antifungal_tablePA %>% select_if(is.numeric) %>% map_dbl(sum) AntiFungal_Richness = as.data.frame(AntiFungal_Richness) AntiFungal_Richness$id <- AntiFungal_Richness$SampleID AntiFungal_Richness = cbind(id = rownames(AntiFungal_Richness), AntiFungal_Richness) rownames(AntiFungal_Richness) <- NULL Metadata_Antifungal_Richness = Metadata %>% left_join(AntiFungal_Richness, by = "id") Metadata_AB = Metadata_Antifungal_Richness %>% left_join(Antifungal_Cultured_Reads3, by = "id") Metadata_AB$metric <- Metadata_AB$TotalAntiFungal_PR * Metadata_AB$Representation write.csv(Metadata_AB, "Metadata_AB.csv", row.names = F) ``` I used the ratio between antifungal matching reads and total cultured reads times the representation of cultured isolates in the sample to have a weighted measure of Bd-inhibitory function, limited by my culturing. I'm confident on my culturing efforts since some samples have over 80% reads represented in my cultured database. This is my data processing. Now the analysis: ###Analysis ##Survival To perform the survival analysis and run the proportional hazards model, I need to have the table organized by frog ID instead of sample. So I did this: ```{r byfrogdata} library(tidyverse) # Load metadata_ab ab <- read.csv("/Users/lauraschuck/Desktop/submissions/antifungal_predictions/Metadata_AB.csv") # Columns to reshape (UPDATED) cols <- c( "Load", "AntiFungal_Richness", "TotalCultured_PR", "TotalAntiFungal_PR", "RatioAF_Full_PR", "Representation", "metric" ) # Confirm colnames names(ab) # Pivot to wide format ab_wide <- ab %>% select(frog_id, time, all_of(cols)) %>% pivot_wider( id_cols = frog_id, names_from = time, values_from = all_of(cols), names_sep = "_" ) # Primeiro, garantir que todas as colunas de load sejam numéricas ab_wide <- ab_wide %>% mutate(across(starts_with("Load_"), ~ as.numeric(.))) # Função para pegar o último valor não-NA last_non_na <- function(x) { rev(x)[which.max(!is.na(rev(x)))] } # Identificar automaticamente todas as colunas de Load_* na ordem correta load_cols <- c( "Load_field", "Load_post_ant", "Load_post_bd", "Load_final", "Load_post_mortem" ) # Selecionar apenas as colunas que realmente existem na sua tabela load_cols <- load_cols[load_cols %in% names(ab_wide)] # Criar load_last ab_wide$load_last <- apply(ab_wide[, load_cols], 1, last_non_na) # Criar shift ab_wide$Load_Shift <- ab_wide$load_last - ab_wide$Load_field ab_wide$Load_Shift_log <- log10(ab_wide$Load_Shift+1) last_info <- ab %>% group_by(frog_id) %>% filter(day == max(day, na.rm = TRUE)) %>% # pega o maior dia slice(1) %>% # garante uma linha ungroup() %>% select(frog_id, day_last = day, status_last = status, species, treatment, ant, bd) ab_wide <- ab_wide %>% left_join(last_info, by = "frog_id") # View result ab_wide write.csv(ab_wide, "metadata_frogid.csv", row.names = FALSE) ``` I created survival curves on JMP: ```{text JMP_survival_script} Local Here( table1 = New Table("metadata_frogid", Add Rows(79), New Script("Source", Open( "$DESKTOP/submissions/metadata_frogid.csv", columns( New Column("frog_id", Numeric, "Continuous", Format("Best", 12)), New Column("Load_post_ant", Numeric, "Continuous", Format("Best", 12)), New Column("Load_post_bd", Numeric, "Continuous", Format("Best", 12)), New Column("Load_final", Numeric, "Continuous", Format("Best", 12)), New Column("Load_post_mortem", Numeric, "Continuous", Format("Best", 12)), New Column("Load_field", Numeric, "Continuous", Format("Best", 12)), ... your remaining columns here ... ), Import Settings( End Of Line(CRLF, CR, LF), End Of Field(Comma, CSV(1)), ... ) ) ), ... more New Column() statements ... ) << Run Formulas; New Window( "metadata_frogid - Survival", table1 << Survival( Y(:day_last), Censor(:status_last), Grouping(:treatment), Censor Code(0), Failure Plot(0), By(:condition) ) ) << Move Window(203, 0) << Set Content Size(573, 973) << Set Window Icon("Survival") ); ``` Now the Proportional Hazards Model, also on JMP: ``` {text JMP_hazard} Set Window Icon( "Model" ); New Window( "metadata_frogid - Fit Proportional Hazards 5", table1 << Fit Model( Y( :day_last ), Effects( :ant, :Load_Shift_log, :Load_Shift_log * :ant, :RatioAF_Full_PR_field ), Personality( "Proportional Hazard" ), Censor( :status_last ), Censor Code( "0" ), Run( Likelihood Confidence Intervals( 1 ), Likelihood Ratio Tests( 1 ) ), By( :species ) ) ``` ##Infection intensity and dispersion overtime ``` {r loads_dispersion} library(nlme) ab <- read.csv("/Users/lauraschuck/Desktop/submissions/antifungal_predictions/Metadata_AB.csv") metadata <- ab metadata$UU_disp <- metadata$UU_dispersion # Make sure 'group' is a factor metadata$group <- as.factor(metadata$frog_id) model1 <- lme( Load_log ~ ant + RatioAF_Full_PR, random = ~ 1 | group, correlation = corCAR1(form = ~ day | group), data = metadata, na.action = na.exclude, method = "REML" ) ##species: metadata_hb <- filter(metadata, species=="Haddadus binotatus") metadata_ih <- filter(metadata, species=="Ischnocnema henselii") model_hb <- lme( Load_log ~ ant + RatioAF_Full_PR, random = ~ 1 | group, correlation = corCAR1(form = ~ day | group), data = metadata_hb, na.action = na.exclude, method = "REML" ) model_ih <- lme( Load_log ~ ant + RatioAF_Full_PR, random = ~ 1 | group, correlation = corCAR1(form = ~ day | group), data = metadata_ih, na.action = na.exclude, method = "REML" ) #check duplicates dups <- metadata[duplicated(metadata[c("frog_id", "day")]), ] print(dups) ##geral model_disp <- lme( UU_disp ~ species + ant + Load_log + ant*Load_log, random = ~ 1 | group, correlation = corCAR1(form = ~ day | group), data = metadata, na.action = na.exclude, method = "REML" ) model_disp_hb <- lme( UU_disp ~ ant + Load_log + ant*Load_log, random = ~ 1 | group, correlation = corCAR1(form = ~ day | group), data = metadata_hb, na.action = na.exclude, method = "REML" ) model_disp_ih <- lme( UU_disp ~ ant + Load_log + ant*Load_log, random = ~ 1 | group, correlation = corCAR1(form = ~ day | group), data = metadata_ih, na.action = na.exclude, method = "REML" ) summary(model_hb) summary(model_ih) summary(model_disp_hb) summary(model_disp_ihsum) ``` ##Lefse ```{r lefse} ## ============================================================= ## LOAD PACKAGES ## ============================================================= library(phyloseq) library(dplyr) library(readr) library(qiime2R) library(ggplot2) library(Biostrings) library(forcats) ## ============================================================= ## IMPORT PHYLOSEQ OBJECT + METADATA ## ============================================================= SVs <- read_qza("table_filt_r5620.qza") taxonomy_qza <- read_qza("dada2_taxonomy.qza") tree <- read_qza("tree_rooted.qza") repseqs <- read_qza("rep-seqs.qza") # <- NOW INCLUDED metadata <- read.csv("Metadata_AB.csv") write.table(metadata, "metadata_final.tsv", sep = "\t", row.names = FALSE, quote = FALSE) physeq <- qza_to_phyloseq( features = "table_filt_r5620.qza", tree = "tree_rooted.qza", taxonomy = "dada2_taxonomy.qza", metadata = "metadata_final.tsv" ) ## ============================================================= ## RUN LEFSE ## ============================================================= lefse <- run_lefse( physeq, group = "species", taxa_rank = "none", kw_cutoff = 0.01, wilcoxon_cutoff = 0.01, lda_cutoff = 3 ) ## ============================================================= ## EXTRACT LEFSE MARKER TABLE CORRECTLY ## ============================================================= mt <- lefse@marker_table@.Data mm_species <- as.data.frame(mt, stringsAsFactors = FALSE) colnames(mm_species) <- lefse@marker_table@names mm_species$Feature.ID <- mm_species$feature ## ============================================================= ## IMPORT TAXONOMY AND MERGE ## ============================================================= tax <- as.data.frame(taxonomy_qza$data) colnames(tax)[1] <- "Feature.ID" lefse_tax <- mm_species %>% left_join(tax, by = "Feature.ID") %>% mutate( last_tax = sapply( strsplit(Taxon, ";"), function(x) trimws(tail(x, 1)) ) ) ## ============================================================= ## LOAD ASV SEQUENCES ## ============================================================= # ASV sequences from rep-seqs.qza asv_seqs <- DNAStringSet(repseqs$data$Sequence) names(asv_seqs) <- repseqs$data$Feature.ID # Subset only ASVs that appear in LEfSe results asv_lefse <- asv_seqs[names(asv_seqs) %in% lefse_tax$Feature.ID] ## ============================================================= ## LOAD INHIBITOR SEQUENCES ## ============================================================= inhib_qza <- read_qza("inhibitor_only_lks.qza")$data inhib_seqs <- DNAStringSet(inhib_qza$Sequence) names(inhib_seqs) <- rownames(inhib_qza) ## ============================================================= ## FUNCTION TO COMPUTE % IDENTITY ## ============================================================= seq_identity <- function(seq1, seq2) { aln <- pairwiseAlignment(seq1, seq2, type = "local") # LOCAL alignment pid(aln) / 100 } ## ============================================================= ## MATCH LEFSE ASVS TO INHIBITOR ASVS AT ≥97% IDENTITY ## ============================================================= identity_results <- sapply(names(asv_lefse), function(asv_id) { seq1 <- asv_lefse[[asv_id]] max(sapply(inhib_seqs, function(seq2) seq_identity(seq1, seq2))) }) # Add match to LEfSe table lefse_tax$inhib_match97 <- lefse_tax$Feature.ID %in% names(identity_results[identity_results >= 0.95]) ## ============================================================= ## BEAUTIFUL LEFSE PLOT WITH STARS FOR INHIBITOR MATCHES ## ============================================================= lefse_tax$label_clean <- fct_reorder(lefse_tax$last_tax, lefse_tax$ef_lda) ggplot(lefse_tax, aes(x = label_clean, y = ef_lda, fill = enrich_group)) + geom_col(width = 0.7, color = "black", linewidth = 0.2) + # STAR FOR INHIBITOR MATCH geom_point( data = subset(lefse_tax, inhib_match97), aes(x = label_clean, y = ef_lda), shape = 8, size = 4, color = "black" ) + coord_flip() + scale_fill_manual( values = c("Haddadus binotatus" = "#F8766D", "Ischnocnema henselii" = "#00BFC4"), name = "Enriched in:" ) + labs( x = "", y = "LDA Effect Size", title = "LEfSe-Enriched ASVs (Stars = ≥97% Similar to Inhibitor Isolates)" ) + theme_minimal(base_size = 14) + theme( plot.title = element_text(face = "bold", size = 18, hjust = 0.5), axis.text.y = element_text(size = 11), axis.text.x = element_text(size = 12), axis.title.x = element_text(size = 14), legend.position = "right", panel.grid.major.y = element_blank(), panel.grid.minor = element_blank() ) ``` ## review ``` {r review} #setwd("C:/Users/laura/iCloudDrive/Desktop/submissions/Animal_microbiome") setwd("/Users/lauraschuck/Desktop/submissions/Animal_microbiome") # ============================================================ # FINAL SURVIVAL ANALYSIS + VISUALIZATION # ============================================================ library(dplyr) library(readr) library(survival) library(survminer) # ----------------------------- # 1) Load and prepare data # ----------------------------- metadata <- read_tsv("metadata_final.tsv", show_col_types = FALSE) metadata_clean <- metadata %>% mutate( frog_id = as.factor(frog_id), species = as.factor(species), ant = as.factor(ant), bd = as.factor(bd), time = as.factor(time), day = as.numeric(day), status = as.numeric(status), Load_log = as.numeric(Load_log), RatioAF_Full_PR = as.numeric(RatioAF_Full_PR) ) # ----------------------------- # 2) One row per frog # ----------------------------- surv_simple <- metadata_clean %>% arrange(frog_id, day) %>% group_by(frog_id) %>% summarise( day_last = max(day, na.rm = TRUE), status_last = max(status, na.rm = TRUE), species = first(species), ant = first(ant), bd = first(bd), RatioAF_Full_PR_field = RatioAF_Full_PR[time == "field"][1], .groups = "drop" ) %>% filter(!is.na(RatioAF_Full_PR_field)) # Relevel for interpretation surv_simple <- surv_simple %>% mutate( species = relevel(species, ref = "Haddadus binotatus"), ant = relevel(ant, ref = "control_a"), bd = relevel(bd, ref = "control_bd") ) # ----------------------------- # 3) GLOBAL MODEL (main result) # ----------------------------- cox_global <- coxph( Surv(day_last, status_last) ~ species + ant * bd + RatioAF_Full_PR_field, data = surv_simple ) summary(cox_global) # ----------------------------- # 4) Kaplan–Meier visualization # ----------------------------- surv_simple$treat <- interaction( surv_simple$species, surv_simple$ant, surv_simple$bd ) fit <- survfit(Surv(day_last, status_last) ~ treat, data = surv_simple) ggsurvplot( fit, data = surv_simple, risk.table = TRUE, pval = TRUE, ggtheme = theme_minimal() ) # ----------------------------- # 5) TIME-VARYING MODEL (mechanism) # ----------------------------- surv_long <- metadata_clean %>% arrange(frog_id, day) %>% group_by(frog_id) %>% mutate( start = day, stop = lead(day), event = lead(status) ) %>% filter(!is.na(stop), stop > start) %>% ungroup() cox_tv <- coxph( Surv(start, stop, event) ~ Load_log + ant + species + cluster(frog_id), data = surv_long ) summary(cox_tv) # H. binotatus: Bd alone vs Bd + antibiotics survdiff(Surv(day_last, status_last) ~ group, data = surv_simple %>% filter(species == "Haddadus binotatus", group %in% c("control_a_bd", "ant_bd"))) # I. henselii: Bd alone vs Bd + antibiotics survdiff(Surv(day_last, status_last) ~ group, data = surv_simple %>% filter(species == "Ischnocnema henselii", group %in% c("control_a_bd", "ant_bd"))) ``` ##ancom ``` {r ancom} # ============================================================ # ANCOM-BC2 at GENUS level — treatment comparison # ============================================================ library(phyloseq) library(ANCOMBC) library(dplyr) library(tibble) ps <- physeq # your object # ------------------------------------------------------------ # 1) Agglomerate to GENUS level # ------------------------------------------------------------ ps_genus <- tax_glom(ps, taxrank = "Genus") # Clean taxa names (optional but helps) tax_table(ps_genus)[, "Genus"] <- make.names(tax_table(ps_genus)[, "Genus"]) # ------------------------------------------------------------ # 2) Use experimental samples (NOT field) # ------------------------------------------------------------ ps_exp <- subset_samples(ps_genus, time != "field") ps_exp <- prune_taxa(taxa_sums(ps_exp) > 0, ps_exp) # Ensure treatment is a factor sample_data(ps_exp)$treatment <- as.factor(sample_data(ps_exp)$treatment) # Check levels (IMPORTANT) levels(sample_data(ps_exp)$treatment) # Set reference level (control) sample_data(ps_exp)$treatment <- relevel( sample_data(ps_exp)$treatment, ref = "control" ) # ------------------------------------------------------------ # 3) Run ANCOM-BC2 # ------------------------------------------------------------ ancom_treat <- ancombc2( data = ps_exp, taxa_are_rows = TRUE, assay_name = "counts", tax_level = "Genus", fix_formula = "treatment", group = "treatment", p_adj_method = "holm", prv_cut = 0.10, lib_cut = 1000, struc_zero = TRUE, neg_lb = TRUE, alpha = 0.05, verbose = TRUE ) # ------------------------------------------------------------ # 4) Extract results # ------------------------------------------------------------ res <- ancom_treat$res # Find columns automatically lfc_cols <- grep("^lfc_treatment", colnames(res), value = TRUE) p_cols <- grep("^p_treatment", colnames(res), value = TRUE) q_cols <- grep("^q_treatment", colnames(res), value = TRUE) diff_cols <- grep("^diff_treatment", colnames(res), value = TRUE) ancom_results <- res %>% as.data.frame() %>% rownames_to_column("Genus") # View top results print(ancom_results, n = 20) # Save write.csv(ancom_results, "ANCOMBC2_treatment_genus.csv", row.names = FALSE) ```