Computational protocol: Links of gut microbiota composition with alcohol dependence syndrome and alcoholic liver disease

Similar protocols

Protocol publication

[…] The analysis of the raw metagenomic reads included quality preprocessing followed by the identification of taxonomic and functional composition (semiquantitative profiling of the abundance of the microbial taxa and functional gene groups, respectively). These analyses were performed as described before [, ], with the modification of the references. The reference sets included a nonredundant set of 353 gut microbial genomes (for the taxonomic analysis) [] and a gut microbial gene catalogue of 9.9 mln genes (for the functional analysis) [].As an additional method for taxonomic profiling, we used MetaPhlAn v2.0 software based on the identification of unique clade-specific genetic sequences []. The read alignment step of MetaPhlAn was performed using Bowtie []. For the MetaFast analysis [], the color-space SOLiD metagenomic reads of the patients and control group were subjected to human sequences filtering, error correction using SAET, and conversion to base-space format. The MetaFast was used with the default settings. The metagenomes were hierarchically clustered using the dissimilarity matrix; the outliers were defined as the metagenomes belonging to the smaller branch after cutting the clustering tree at the top level. [...] Statistical analysis of the microbial compositional data was performed in R []. The pairwise dissimilarity between the community structures was assessed using Bray-Curtis measure. The taxa or genes differentially abundant between the groups of the metagenomes were identified using Mann-Whitney rank test (multiple comparison adjustment using Benjamini-Hochberg (FDR), significance threshold: adjusted p < 0.01). For the comparative analysis, only the genera and species with the abundance of > 1 and > 0.1%, respectively, in at least one of the samples were included.For the analysis of the impact of each clinical factor on the gut community structure, the list of the factors included gender, age, the presence of alcohol dependence, and the presence of liver cirrhosis. Firstly, the effect of each of the factors on the overall community structure was assessed using PERMANOVA with Bray-Curtis dissimilarity metric. Then the resulting factors significantly associated with the taxonomic composition were included into the multifactor analysis using MaAsLin package ( The relative abundance values of the taxa were treated as the dependent variables. The model scheme for each microbial genus/species can be summarized as follows: relative abundance of genus/species~alcoholic dependence + liver cirrhosis + gender For each subject, the factors “alcoholic dependence” and “liver cirrhosis” were assigned the values “yes” or “no” according to the group of the subject: Low-abundant taxa were filtered as described above (the final list of the taxa included 73 genera and 262 species). The MaAsLin method was used with the following parameters: the significance level was 0.05 and the dMinSamp parameter was used to limit the analysis to the taxa that were abundant at the level of > 0.01% in at least 10 samples.The ranks of oral microbial species were compared as follows. Suppose N is the number of the species detected in at least one sample. For each sample, all N species were sorted in the decreasing order of abundance and assigned ranks (1—the most abundant, N—the least abundant; the species not detected in the sample were assigned N + 1). For each species, the respective vectors of microbial ranks were compared using Mann-Whitney test (significance threshold: p < 0.05). […]

Pipeline specifications

Software tools MetaPhlAn, Bowtie, MetaFast, MaAsLin
Application Metagenomic sequencing analysis
Organisms Homo sapiens
Diseases Alcoholism, Fibrosis, Inflammation, Liver Cirrhosis, Liver Diseases, Liver Diseases, Alcoholic, Colorectal Neoplasms, Substance-Related Disorders, Alcohol-Related Disorders, Dysbiosis
Chemicals Acetaldehyde, Ethanol, Butyrates