Review Reports

S/NO

REVIEWER COMMENTS

AUTHOR RESPONSES

(1)

The methods were unclear, authors have described as all samples were pooled into three samples, and how many samples were sequenced? In addition, how did DESeq2 handle only one sample? Or two sample?

Reference has been made to lines 126–131: Samples from Kakamega, Nandi, and Vihiga were combined into a single composite sample labeled RVK1. Similarly, all Bungoma samples were pooled into one composite sample, RBGM1, and those from Busia into RBU1. These three pooled samples were submitted to Novogene Limited Company in Singapore for metatranscriptomic analysis and sequencing. Since only three composite samples were sequenced, the use of DESeq analysis was thus possible.

(2)

Authors have assembled RNA-seq reads, and the unique genes were removed redundancy using CORSET, but there were also 532902 unigenes remaining, the unigenes were so many to analyze in depth, which would be biased in conclusion.

We appreciate the reviewer’s observation. Indeed, a large number of unigenes (532,902) were generated after assembly and redundancy removal using CORSET. This high number reflects the complexity and diversity of the transcriptomic data, especially considering that the samples originated from multiple regions with potentially diverse microbial communities and host responses. While it was not feasible to analyze each unigene in depth, we focused our functional analysis on annotated and functionally relevant unigenes through alignment to curated databases (e.g., KEGG, eggNOG, CAZy). This approach allowed us to extract meaningful biological insights while minimizing bias

(3)

The figure 2 would been selected from advertisement of Novogene Limited Company, which could be deleted.

Figure 2 has been removed from the manuscript in accordance with the reviewer’s comments.

(4)

Table 2, 3, 4, and 5 contained low information content, and they could be deleted. Especially did table 5.

Tables 2, 3, 4, and 5 have been removed from the main manuscript and relocated to Supplementary Material 4 for reference.

(5) .

Figure 3 could adopt logarithmic coordinates, which would be beautiful. Similar, many genes expressional levels would be displayed with log(expression values) for better understanding

The authors believe that Figure 3 effectively conveys their findings and remains appropriate for inclusion.

(6)

I didn’t know which means for three samples with PCA analysis, see figure 6.

Principal Component Analysis (PCA) was conducted using functional abundance data from various databases across different classification levels. In the resulting PCA plots, samples with more similar functional profiles appeared closer together. The analysis, based on functional abundance at KEGG Level 2, Level 3, and KO levels, revealed significant differences among the samples (see lines 294–298).

(7) ?

From figure 9, there were many unigenes identified as DEGs, but many DEGs have similar expressional patterns, were these unigenes correctly identified

The differentially expressed genes (DEGs) shown in Figure 9 were identified using standard statistical thresholds (e.g., adjusted p-value < 0.05 and |log2 fold change| ≥ 1) through DESeq2 analysis. While many DEGs exhibit similar expression patterns, this is expected in cases where genes may be co-regulated or functionally related, particularly under similar biological or environmental conditions. To ensure accuracy, we cross-referenced DEG annotations with functional databases, and included only those with significant and consistent expression patterns across replicates.

(8)

All analysis were performed based RNA-seq analysis, there was no molecular experiment to support their results, such as qRT-PCR, etc, please provide more evidence about their conclusion.

qRT-PCR was not performed to validate the molecular findings, as noted. However, this has been acknowledged and documented as a limitation of the study (see lines 853–855).

(9)

Many figures in the manuscript could be merged. It was recommended to combine them.

Most of the figures in the manuscript are already consolidated to effectively present the data visually. Additionally, the removal of Figure 2 has further reduced the total number of figures.

(10)

The conclusion section was too long, please streamline it. In addition, was the Recommendations section needed for this manuscript?

The conclusion may be lengthy, but it presents valuable insights derived from our findings. Similarly, the recommendations address key knowledge gaps identified during the course of the study.

S/NO

REVIEWER COMMENTS

AUTHOR RESPONSES

(1)

The methods were unclear, authors have described as all samples were pooled into three samples, and how many samples were sequenced? In addition, how did DESeq2 handle only one sample? Or two sample?

Reference has been made to lines 126–131: Samples from Kakamega, Nandi, and Vihiga were combined into a single composite sample labeled RVK1. Similarly, all Bungoma samples were pooled into one composite sample, RBGM1, and those from Busia into RBU1. These three pooled samples were submitted to Novogene Limited Company in Singapore for metatranscriptomic analysis and sequencing. Since only three composite samples were sequenced, the use of DESeq analysis was thus possible.

(2)

Authors have assembled RNA-seq reads, and the unique genes were removed redundancy using CORSET, but there were also 532902 unigenes remaining, the unigenes were so many to analyze in depth, which would be biased in conclusion.

We appreciate the reviewer’s observation. Indeed, a large number of unigenes (532,902) were generated after assembly and redundancy removal using CORSET. This high number reflects the complexity and diversity of the transcriptomic data, especially considering that the samples originated from multiple regions with potentially diverse microbial communities and host responses. While it was not feasible to analyze each unigene in depth, we focused our functional analysis on annotated and functionally relevant unigenes through alignment to curated databases (e.g., KEGG, eggNOG, CAZy). This approach allowed us to extract meaningful biological insights while minimizing bias

(3)

The figure 2 would been selected from advertisement of Novogene Limited Company, which could be deleted.

Figure 2 has been removed from the manuscript in accordance with the reviewer’s comments.

(4)

Table 2, 3, 4, and 5 contained low information content, and they could be deleted. Especially did table 5.

Tables 2, 3, 4, and 5 have been removed from the main manuscript and relocated to Supplementary Material 4 for reference.

(5) .

Figure 3 could adopt logarithmic coordinates, which would be beautiful. Similar, many genes expressional levels would be displayed with log(expression values) for better understanding

The authors believe that Figure 3 effectively conveys their findings and remains appropriate for inclusion.

(6)

I didn’t know which means for three samples with PCA analysis, see figure 6.

Principal Component Analysis (PCA) was conducted using functional abundance data from various databases across different classification levels. In the resulting PCA plots, samples with more similar functional profiles appeared closer together. The analysis, based on functional abundance at KEGG Level 2, Level 3, and KO levels, revealed significant differences among the samples (see lines 294–298).

(7) ?

From figure 9, there were many unigenes identified as DEGs, but many DEGs have similar expressional patterns, were these unigenes correctly identified

The differentially expressed genes (DEGs) shown in Figure 9 were identified using standard statistical thresholds (e.g., adjusted p-value < 0.05 and |log2 fold change| ≥ 1) through DESeq2 analysis. While many DEGs exhibit similar expression patterns, this is expected in cases where genes may be co-regulated or functionally related, particularly under similar biological or environmental conditions. To ensure accuracy, we cross-referenced DEG annotations with functional databases, and included only those with significant and consistent expression patterns across replicates.

(8)

All analysis were performed based RNA-seq analysis, there was no molecular experiment to support their results, such as qRT-PCR, etc, please provide more evidence about their conclusion.

qRT-PCR was not performed to validate the molecular findings, as noted. However, this has been acknowledged and documented as a limitation of the study (see lines 853–855).

(9)

Many figures in the manuscript could be merged. It was recommended to combine them.

Most of the figures in the manuscript are already consolidated to effectively present the data visually. Additionally, the removal of Figure 2 has further reduced the total number of figures.

(10)

The conclusion section was too long, please streamline it. In addition, was the Recommendations section needed for this manuscript?

The conclusion may be lengthy, but it presents valuable insights derived from our findings. Similarly, the recommendations address key knowledge gaps identified during the course of the study.

Reviewers Concers

Authors response

1

For example, as authors described, there were on replacements for each sample, how to perform DEGs analysis using DESeq2? Which need repeats in analysis process.

We have taken time to carefully re-examine our analytical approach in light of the number of samples used and as guided by the reviewer. Accordingly, we have performed a differential expression analysis using  EdgeR package, with the TMM normalization method to curate this error. Our decision to employ EdgeR was because it applies a Poisson distribution model to evaluate differences in gene expression. In view of the foregoing, all references to DESeq2 have been collapsed from the methods section of the manuscript. This fundamentally eliminated the bias. See 180-183.

2

In addition, in comments, comments #5, the logarithmic coordinates would be more beautiful for displaying, but authors didn’t adopt, which made this? Authors have no capability for re-plotting this figure with logarithmic coordinate? All authors could not reshape their data? Authors have described as

As already acknowledged in our previous response, the Corset software was used in transcript Redundancy Removal and Length Distribution. To reduce transcript redundancy and improve clustering efficiency, transcript-level counts were clustered using the Corset tool. Corset clusters transcripts based on shared read support and expression across samples, which collapses isoforms and redundant sequences. This allows for more accurate downstream differential expression analysis by focusing on gene-level signal rather than highly redundant transcript variants.

In addition, the distribution of transcript lengths of the clustered transcripts was verified to ensure the quality and representativeness of the dataset. The clusters exhibited a broad transcript length range, in agreement with expected biological diversity, and indicative of efficient redundancy removal.

Overview of the transcripts and unigenes

The total number of unigenes (the longest transcript of each gene) based on the length distribution was 532902, and it is these unigenes that were used for downstream analysis.

The authors still maintain that Figure 2 accurately presents the gene ontology findings. This fact is corroborated by previous published works (Martin et al., 2016), (Kundu et al., 2019)