Download: pairseq-data.tgz

These directories contain the data and results of the analyses reported in "High-Throughput Pairing of T Cell Receptor Alpha and Beta Sequences", Howie et. al, 2015, Science Translational Medicine.

The data produced by the pairSEQ assay are in directories named "experimentN/cdna_data/", for N in (1,2,3,4,5), where the experiment numbers correspond to the labels used in the manuscript. Each of these directories contains 192 files: 96 for TCRA and 96 for TCRB, numbered consecutively to denote different wells on a 96-well plate. After DNA sequencing, clone sequences were mapped to wells of origin via well-specific barcodes that were incorporated into the cDNA molecules; each file contains sequences that mapped to a given well. The files are in "results.tsv" format, which includes a column of clone sequences (87 bp that span the CDR3) sorted by read count ("copy" column) and followed by sequence annotations. Key annotations include V- and J-gene assignments and "sequenceStatus" ("In" implies that a receptor sequence is in-frame and likely to be productively expressed).

Each "experimentN/" directory also contains a file named "tcr_pairseq_fdr1pct.pairs". These files show the results of our pairing and false discovery rate analyses: for each experiment, we report a list of pairs that were called with FDR <= 1%. Each pairs file includes six columns: TCRA sequence, list of wells occupied by TCRA (comma-separated), TCRB sequence, list of wells occupied by TCRB (comma-separated), Jaccard index (a measure of well occupancy overlap), and p-value (a measurement of how likely it would be for the observed occupancy overlap to occur by chance).

In each of Experiments 1 and 2, two blood samples were multiplexed on a single plate, which was then processed with the pairSEQ assay. To demultiplex the samples, we performed deep repertoire sequencing of the separate donors who provided the samples ("Subject X" and "Subject Y"). The repertoire sequencing data are provided in the "subjectX/" and "subjectY/" directories. Each of these folders contains subdirectories named "cdna_data" and "gdna_data". For cDNA repertoire sequencing, we used 96 sequencing reactions per locus, so there are 192 files per subject. For gDNA repertoire sequencing, we used 48 reactions per locus, so there are 96 files per subject. The gDNA data were used for repertoire frequency measurements in Figure 4 and sample-of-origin assignments in Figure 3; the cDNA data were used for sample-of-origin assignments in Figure S2.

As in Experiments 1 and 2, Experiments 4 and 5 used plates with multiplexed samples: nine tumor samples and nine blood samples, respectively. We separately sequenced gDNA from each of these samples to facilitate demultiplexing; the gDNA sequences can be found in "experiment4/gdna_data/" and "experiment5/gdna_data/", where they are subdivided into sample-specific directories (four from patients with breast tumors, four from patients with kidney tumors, and one from a patient with a lung tumor).