This repository contains code used to reanalyse Human Microbiome Project data to benchmark MAPseq.
Raw data from the HMP is pre-processed and matched to samples and sequencing subregions in the script prepare.raw_data.pl. This generates one folder per sample for subsequent sample-wise (parallel) processing. It also generates global mapping tables.
The script submit.filter_chimeras.sh then calls the script filter.chimeric_reads.sh for each sample in parallel to remove chimeras.
The script submit.align.hmp_samples.sh calls the script align.hmp_samples.sh for each sample in parallel to run INFERNAL. Afterwards, the script make_alignments.global.sh stitches together individual per-sample alignments and de-replicates and de-noises them a bit. This script also defines the "consensus" lists of sequences to be used downstream by clustering methods – keeping only those sequences which are non-chimeric and align satisfactorily to the target subregion.
The scripts "make_otus.*.sh" then call the different mapping/clustering tools to generate OTU sets and translate them into R-readable formats (OTU tables etc.).
Finally, the script hmp.benchmark.R contains all the code for R-based analyses, as detailed in the manuscript.
(Raw) results are available in the folder results.