gregcaporaso
commented
11 years ago If it helps, here are the steps that I ran for the Student Microbiome Project analysis:
echo "filter_otus_from_otu_table.py -i /Users/caporaso/analysis/student-microbiome-project/otu_tables/closed_ref_otu_table.biom -o /Users/caporaso/analysis/student-microbiome-project/mislabeling/closed_ref_otu_table_s10.biom -s 10" | qsub -keo -N filtotus -l pvmem=64gb -q memroute
echo "single_rarefaction.py -i /Users/caporaso/analysis/student-microbiome-project/mislabeling/closed_ref_otu_table_s10.biom -o /Users/caporaso/analysis/student-microbiome-project/mislabeling/closed_ref_otu_table_s10_1000.biom -d 1000" | qsub -k eo -N smp-rare -l pvmem=64gb -q memroute
filter_samples_from_otu_table.py -i /Users/caporaso/analysis/student-microbiome-project/mislabeling/closed_ref_otu_table_s10_1000.biom -o /Users/caporaso/analysis/student-microbiome-project/mislabeling/closed_ref_otu_table_s10_1000_no_controls.biom -m /Users/caporaso/analysis/student-microbiome-project/StudentMicrobiomeProject-map.tsv -s 'Control:no'
echo "supervised_learning.py -i /Users/caporaso/analysis/student-microbiome-project/mislabeling/closed_ref_otu_table_s10_1000_no_controls.biom -m /Users/caporaso/analysis/student-microbiome-project/StudentMicrobiomeProject-map.tsv -c "BodySite" -e cv5 -o /Users/caporaso/analysis/student-microbiome-project/mislabeling/s10_1000/" | qsub -keo -N smp_rf -l pvmem=16gb -q memroute
cd /Users/caporaso/analysis/student-microbiome-project/mislabeling/s10_1000/
make_3d_plots.py -i ../../bdiv_even10000/unweighted_unifrac_pc.txt -m mislabeling.txt -o unifrac_color_by_mislabeling -b mislabeled_at_0.05,mislabeled_at_0.10,mislabeled_at_0.15,mislabeled_at_0.20,mislabeled_at_0.25,mislabeled_at_0.30,mislabeled_at_0.35,mislabeled_at_0.40,mislabeled_at_0.45,mislabeled_at_0.50
provided more detailed documentation on how to identify and remove mislabeled samples from an OTU table. This should include an illustration of how to create a 3D PCoA plot where samples are colored by their probability of being mislabeled