ekopylova/otu-clustering

Scripts for benchmarking and comparison of short read OTU clustering tools available in QIIME

OTU-clustering

Scripts for benchmarking and comparison of short read OTU clustering tools available via QIIME 1.9.0. The full benchmark can be launched with the script OTU-clustering/shell_scripts/launch_benchmark.sh. This script will execute all software tools and perform analysis using the datasets & scripts below. The user must modify the working directory path in launch_benchmark.sh prior to executing this script. Only commands for launching software in OTU-clustering/shell_scripts/commands_16S.sh and OTU-clustering/shell_scripts/commands_18S.sh are called using the qsub environment, although this is easily modifiable to run on any system.

Dependencies

1. QIIME 1.9

Simulate reads:

2. PrimerProspector (1.0.1)
3. ART (VanillaIceCream-03-11-2014)

Benchmark:

4. BLAST 2.2.29+
5. USEARCH Uchime (7.0.1090)
6. UPARSE (7.0.1090)

Install

git clone https://github.com/ekopylova/OTU-clustering.git

Datasets

1. Read datasets: ftp://ftp.microbio.me/pub/supplemental_otu_clustering_datasets.tar.gz
2. Greengenes 13.8: ftp://ftp.greengenes.microbio.me/greengenes_release/gg_13_5/gg_13_8_otus.tar.gz
3. SILVA 111: ftp://ftp.microbio.me/pub/QIIME_nonstandard_referencedb/Silva_111.tgz
4. 16S PyNAST template: http://greengenes.lbl.gov/Download/Sequence_Data/Fasta_data_files/core_set_aligned.fasta.imputed
5. 18S PyNAST template: ftp://ftp.microbio.me/pub/core_Silva119_alignment.fna.gz
6. For chimera checking 16S, the gold database: http://drive5.com/uchime/uchime_download.html
7. For chimera checking 18S, the SILVA 97% representative set from SILVA 111 (see 3)

Scripts

The benchmarking and analysis comparison can be executed using the following scripts (in given order). Scripts 2-13 require input arguments, all of which are defined in launch_benchmark.sh.

1. Launch full benchmark (executes all scripts below):
   OTU-clustering/shell_scripts/launch_benchmark.sh
   
   

Otherwise, the user may launch each script individually,

2. Simulate even and staggered community reads:
   OTU-clustering/shell_scripts/simulate_reads.sh
3. Launch all software (via QIIME’s pick_closed_reference_otus.py, pick_de_novo_otus.py and pick_open_reference_otus.py) on 16S datasets:
   OTU-clustering/shell_scripts/commands_16S.sh
4. Launch all software on 18S datasets:
   OTU-clustering/shell_scripts/commands_18S.sh
5. Remove singleton OTUs (OTUs consisting of only 1 read) from the final OTU tables generated in steps 2 and 3:
   OTU-clustering/python_scripts/run_filter_singleton_otus.py
6. Summarize taxonomy using filtered OTU tables:
   OTU-clustering/python_scripts/run_summarize_taxa.py
7. Summarize filtered OTU tables:
   OTU-clustering/python_scripts/run_summarize_tables.py
8. Compute true positive, false positive, false negative, precision, recall, F-measure and FP-chimera, FP-known, FP-other metrics using the summarized taxonomy results:
   OTU-clustering/python_scripts/run_compute_precision_recall.py
9. Generate alpha diversity plots:
   OTU-clustering/python_scripts/run_single_rarefaction_and_plot.py
10. Generate beta diversity plots:
    OTU-clustering/python_scripts/run_beta_diversity_and_procrustes.py
11. Generate taxonomy comparison tables:
    OTU-clustering/python_scripts/run_compare_taxa_summaries.py
12. Generate taxonomy stacked bar plots:
    OTU-clustering/python_scripts/run_generate_taxa_barcharts.py
13. Plot TP, FP-chimera, FP-known and FP-other results:
    OTU-clustering/python_scripts/plot_tp_fp_distribution.py

Citing

If you use any of the data or code included in this repository, please cite with the URL: https://github.com/ekopylova/OTU-clustering.git