Wound image and transcriptome datasets of swine acute wounds

Wound healing progresses through overlapping phases: hemostasis, inflammation, proliferation, and remodeling. Continuous characterization of these transitions remains limited. Here, we employed a swine excisional wound model to monitor cellular dynamics across the healing timeline. Both non-invasive imaging and wound biopsy samples from the wound edge and center were acquired. Wound photographs were analyzed using advanced artificial intelligence methods. Wound biopsy samples were subject to RNA sequencing to generate gene expression profiles for the course of healing. By combining the image and the gene expression analyses, we were able to create the comprehensive data for wound healing, which can serve as ground truth for building wound diagnostic and treatment algorithms.

Methods
Six domestic pigs (Yorkshire-mix breed, females, 45-50 Kg) were utilized and divided to 2 groups (3 animals/group) for wound biopsy collection in a rotation order (group 1 biopsy on post-op day 1, 3, 5, 7, 11, 15, and 21/endpoint; group 2 biopsy on day 2, 4, 6, 9, 13, 16, and 19/endpoint).  Twelve full-thickness, excisional wounds at 2cm in diameter were created bilaterally on each side of the dorsum of each animal after shaving, depilation, and skin preparation.  Wound images (day 0-21) were captured with a DSLR camera at a set distance of 1 foot above the wounds, and a photo scale (Medline NE1 Wound Assessment Tool) was used for calibration in each image.  To determine the wound healing timeline, open wound areas were analyzed by manual tracking in ImageJ or automatically cropped and analyzed by algorithms.

RNA samples (a total of 150 samples of 72 paired samples from the wound edge and center, and 6 healthy skin samples as controls) were extracted from the wound biopsies with Qiagen RNeasy Fibrous Tissue Mini Kit (Qiagen, German).  Around 200ng mRNA/sample were sent to Novogene (Beijiang, China) for RNAseq analysis, including RNA sample quality control, library preparation with poly-A enrichment, sequencing by Illumina sequencing system NovaSeq X Plus (PE150) for >20 millions paired reads, generation of raw data in the FASTQ format, and aligned for gene counts.  

The transcriptome data is represented by the table of gene expression counts "gene_count.xlsx", where each row represents a gene and each column represents a sample.  The column names are SampleID labels.  The encoding of sample labels is explained in the table "CodesDaVinci.xlsx".  This table contains the pig ID, wound number, day of collection from the wound onset, and wound location label (center or edge). The expression time series plots of other genes can be found through an online tool at https://geneexpressionsearch.pythonanywhere.com.