human-microbiome-project-dataset/s3://human-microbiome-project/HHS/HMASM

In the first phase of WGS sequencing, 752 samples were sequenced, comprising
15 body sites. Of these, 690 passed our quality control screens, which
included identification of outliers by mean contig & ORF density, human hits,
rRNA hits and size. Reads and assemblies are available here for these 690
samples.

This dataset includes 35.85 billion human contaminant-screened reads in FASTQ
format, which are 2.3 TB in size, compressed. Reads from each individual
sample were assembled using SOAP, generating 48.3 million scaffolds with a
total compressed size of 13 GB.

WGS reads are referred to as "WGS". Assemblies are referred to as "PGA". Files
are grouped first by body site, and then by sample identifier for download of
WGS and PGA files.