reAnalyze #1 - Skin Disease
Last updated
Last updated
Recent advancements in sequencing technology and database development have provided new insights into microbial communities. Notably, many microbial genomes have been sequenced from metagenomes without official descriptions. With up-to-date databases, especially those incorporating genomospecies, and improved software, we can revisit existing datasets to uncover previously overlooked organisms.
Our goal with these reAnalyze blog posts is to utilize EzBioCloud’s comprehensive databases and proprietary pipeline to explore public metagenomic datasets and share relevant findings.
Let's reAnalyze: Unexplored diversity and strain-level structure of the skin microbiome associated with psoriasis by Tett et al., 2017. This paper we are looking at today considered the differences between healthy and psoriasis skin microbiomes. They sampled from elbows and behind ears, using clear skin sites from the same individual as controls. The sequencing data are available publicly, with some missing samples, under the BioProject ID: PRJNA281366.
For microbial profiling, they used MetaPhlAn 2.0, released in 2015. Although MetaPhlAn 2.0 was advanced for its time, MetaPhlAn 4.0, with approximately six times more microbial clades in its reference database, offers updated nomenclature and taxonomy. This highlights the importance of revisiting older datasets with modern tools.
When we think of the microbiome, our minds often turn to the gut. However, the largest organ of our body, the skin, is often overlooked despite it being the interface between our bodies and the environment. The skin's microbiome is a critical component of our overall health, interacting continuously with external factors and reflecting our personal ‘exposome’ - the sum of all environmental exposures we encounter throughout our lives.
The skin's biogeography, influenced by physiological factors such as sebaceous, moist, or dry regions, contributes to the dynamic microbial community on our skin (Oh et al., 2016). These regions create unique microenvironments that support diverse microbial populations (Yang et al., 2022). Moreover, individual attributes and personal care routines further contribute to the skin's microbial diversity, resulting in almost personal signatures that are unique to each person and specific to different body regions.
In our analysis, the alpha diversity of each cohort showed similar patterns to the paper: across all metrics, healthy skin samples consistently show higher diversity and evenness compared to diseased skin samples. Diseased skin tends to have lower species richness, evenness, and diversity, indicating a microbial community shift that could be associated with disease states (Figure 1).
A closer examination can reveal which species might be responsible for these differences and whether MetaPhlAn 2.0 missed any significant taxa so let’s run the ‘Group Composition’ and ‘Differential Abundance Analysis’ tools on this dataset.
The most common microbial species across skin sites irrespective of psoriasis were Staphylococcus epidermidis, Cutibacterium acnes (in the paper referred to as its basonym: Propionibacterium acnes), Staphylococcus capitis, and Micrococcus luteus.
“Culture dependent studies found these were associated with disease (psoriasis as a skin disease) exacerbation: Staphylococcus aureus, Streptococcus pyogenes, and fungi such as Malassezia.”
The potentially responsible species can already be seen in the Group Composition plot of this dataset (Figure 2). Healthy skin samples are dominated by Staphylococcus epidermidis, a common skin commensal. Whereas diseased skin samples show a shift towards potentially pathogenic taxa such as Staphylococcus aureus (the red bars, also noted in the original paper) and a few Corynebacterium species, which may be associated with the disease state. Then let’s test if any of these are statistically abundant in the diseased dataset samples.
Differential abundance analyses identify taxa that are absent in one cohort or found in significantly increased abundances in another. For example, certain species were consistently present in diseased sites but not in healthy ones, and deeper analysis revealed strain differences between diseased and healthy sites. Tools like LEfSe (Linear Discriminant Analysis Effect Size) are commonly used for such differential abundance analyses. The resulting taxa are potential biomarker candidates.
The notorious Staphylococcus aureus was found to be differentially abundant by ANCOMBC and MaAsLin2 in the diseased samples (Figure 3a). This species is a known pathogen with clinical relevance (Figure 4). LEfSe found nine other lesser-known species. Interestingly, there appeared to be far more species associated with healthy skin samples than diseased. Most notably, Cutibacterium acnes and Staphylococcus epidermidis were more abundant in the healthy samples which reached a significant consensus among six of the tools (Figure 3b).
Their results indicated no clear biomarkers associated with psoriatic lesion diseases at a species level, even if there were varying abundances. Considering there are a few dominant species in both cohorts, they decided to look closer at the strain level to see if they could differentiate the species between diseased and affected. They achieved this with StrainPhlAn which uses taxon-specific SNPs of S. epidermidis and P. acnes. They complimented this with metaMLST. The results show some heterogeneity at the strain-level within and between patients even virulence genes from commensal strains only in diseased sites however, the results were not definitive. As per usual, a larger sample size and deeper sequencing (difficult when it comes to skin samples) would provide better results so keep on sequencing and sharing.
Revisiting this dataset through reAnalyze, we identified that Common species like Staphylococcus epidermidis and Cutibacterium acnes dominate healthy skin, while diseased skin shows an increase in potentially pathogenic taxa such as Staphylococcus aureus. Differential abundance analysis identified Staphylococcus aureus as significantly more abundant in diseased samples, while Staphylococcus epidermidis and Cutibacterium acnes were more abundant in healthy samples (they used these two species in their strain analysis), highlighting them as potential biomarkers for skin health. LEfSe provided the most interesting result with nine species associated with psoriatic lesion microbiomes. Although, the other differential abundance analysis tools didn’t register them, they are still candidate biomarkers to explore more.
The comparison of metagenomic cohorts using differential abundance analyses provides valuable insights into the microbial dynamics of healthy and diseased skin. By revisiting older datasets with updated tools and databases, we can uncover new findings that enhance our understanding of the skin microbiome and its role in health and disease. This is our intention with reAnalyze.
Oh, J., Byrd, A. L., Park, M., Kong, H. H., & Segre, J. A. (2016). Temporal stability of the human skin microbiome. Cell, 165(4), 854-866.
Tett, A., Pasolli, E., Farina, S., Truong, D. T., Asnicar, F., Zolfo, M., ... & Segata, N. (2017). Unexplored diversity and strain-level structure of the skin microbiome associated with psoriasis. NPJ biofilms and microbiomes, 3(1), 14.
Yang, Y., Qu, L., Mijakovic, I., & Wei, Y. (2022). Advances in the human skin microbiota and its roles in cutaneous diseases. Microbial Cell Factories, 21(1), 176.
