2.4. Single-Nucleotide Mutation Annotation and Statistics Analysis

The Spearman correlation between microbial species and Lp082 was calculated using the “WGCNA” package (version 1.73) [23] in R software (version 4.3.2) [24]. Species were screened based on correlation results with a p-value < 0.05 and correlation coefficients with Lp082 either greater than 0.5 or less than −0.5. Genomes of representative strains of these species were collected. The FASTA files of representative strain sequences with a “Reference genomes” were downloaded from the National Center for Biotechnology Information (NCBI) (Table S1), and a representative strains database was constructed using Bowtie2 [25]. The metagenomic sequencing reads from both intestinal segments at each time point were aligned against the reference database to obtain Binary Alignment/Map (BAM) files for the intestinal content samples. Mutation sites were annotated using inStrain v1.0.0 software to obtain SNV information between intestinal strains and the reference genome [26]. The command is as follows: prodigal-i/home/ref.fna-d ref_genes.fna, bowtie2-build/home/inStrain/ref.fna ref.index; bowtie2-p 36-x/home/inStrain/ref.index --no-mixed --very-sensitive --n-ceil 0,0.01-U/home/A.fq.gz-S A.sam; samtools view-bS/home/inStrain/A.sam-oA.bam; samtools sort-m 10000000000/home/inStrain/A.bam A.sorted.bam; inStrain profile/home/inStrain/A.sorted.bam/home/inStrain/ref.fna-c 100-f 0.49-o A.profile-p 36-g/home/inStrain/ref_genes.fna. Using a control group as a negative control, we excluded SNVs arising from issues related to the selection of the reference genome. The SNVs detected in both groups from the same day’s samples marked for natural variation were generated over time and subsequently removed. Ultimately, only the SNVs induced by probiotic intervention in the local gut microbiota were retained. The *.SNVs.tsv files in the output folder include details such as strain codes, mutation positions, the bases before and after the mutation, and whether the mutations are synonymous or non-synonymous, and so on.

Analysis of Variance (ANOVA) [27] was applied for comparisons among multiple groups, followed by Tukey’s test for pairwise comparisons [28]. When data did not meet the assumptions of homogeneity of variance or normal distribution, the Kruskal–Wallis test was used [29]. Statistical significance was assessed using p = 0.05. Utilizing the R packages “vegan” (version 2.6.8) [30], we computed the α-diversity indices Shannon and Simpson and subsequently produced boxplots (mean and SD) through the “ggplot2” package (version 3.5.0) [31]. To further verify the statistical significance of these differences, the Kruskal–Wallis test from the “permute” package (version 0.9.7) [32] was employed. β diversity based on Bray–Curtis [33] distance was calculated using the “vegan” package and subsequently visualized as a PCoA [34] plot with “ggplot2”. Box plots depicting species abundance were constructed using GraphPad Prism software (version 8.0.2) [35]. Network diagrams were then constructed using Cytoscape software (version 3.10.1) [36]. Genomic General feature format (GFF) files for the species were downloaded from NCBI to annotate gene functions. Protein prediction maps for relevant genes were generated using Protein Homology/analogY Recognition Engine V 2.0 [37].