Mining Candidate Genes and Favorable Haplotypes for Flag Leaf Shape in Rice (Oryza sativa L.) Based on a Genome-Wide Association Study

Round 1

Reviewer 1 Report

The Manuscript “Candidate gene mining and favorable haplotypes for flag leaf shapes in rice (Oryza sativa L.) based on genome-wide association study” aimed to detect association signals between flag leaf shape-related traits and nucleotide variants in 173 rice panels. Authors investigated four flag leaf traits, such as FLL, FLW, FLR, and FLA in two trials to obtain phenotype data. NGS for 173 accessions was conducted and total 1,326,094 SNPs were finally detected. Association analyses were performed by MLM and total 11 significant signals for four flag leaf traits were simultaneously detected for two years. Three candidate genes were suggested and were used for haplotype analysis.      

 The topic is important in rice breeding to improve plant architecture as flag leaf shape is one of the major components which affects photosynthesis, biomass, and grain yield. This study is excellent in that it measures repetitively phenotypes for two years and uses a new association panel. However, this manuscript should be improved overall. Description in the MS was inadequate and the data did not match the description. MS also has some statistical issues and arbitrary or over-interpretation of the results. In addition, detailed information should be described in the material and method section. This manuscript needs a major revision for publication. 

The additional comments are as follows:

1. A basis for the formula ‘FLA=FLL×FLW×0.75’ should be provided. Why was 0.75 multiplied?

2. Methods for genomic analyses, including NGS library, align tools, parameters and gene annotation database (MSU) used in this study should be presented.

3. Please explain what covariates were used for MLM and how covariates were estimated.

4. If q-value estimated from BH-FDR correction had been used for deciding association thresholds, each p-value would have been different for each phenotype. As authors used p = 1.0 × 10-5 as the fixed threshold, the description that threshold was established by FDR is inappropriate. Moreover, to detect significantly associated SNPs without false-positive error, stricter thresholds should be used for association analysis by a million markers.

5. The explanation for determining candidate genes near lead SNPs is not clear. It is not based on LD decay as well as LD block. Even LD block is not presented in LD heatmap in Figure 6.

6. Results 3.2 should be modified overall. Description for parameters, including nucleotide diversity, SNP density, and others, is wrong and parameter-related data was not presented in MS.

7. Comparison of phenotypes among haplotypes should be performed using ANOVA and multiple comparisons.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

In the MS "Candidate gene mining and favorable haplotypes for flag leaf 2 shapes in rice (Oryza sativa L.) based on genome-wide associa-3 tion study", authors investigated genes related to several shape of rice flag leaf by GWAS approach.

While the experimental design is clear and adeguate, the MS must be carefully revised for english style and punctuation. Please, find attached the revised MS, correlated with comments and several paragraph highlighted for english style check and revision.

At the present stage, the MS development requires a good revision work, as in the present form it seems to me sloopy developed (see hihglighted paragraphs and notice mixed verb tenses and short sentences appearing like "little-thoughts".

For these regions, I evaluated this manuscript as reconsidered after major revisions.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The questions and concerns have been well addressed. There are now remaining minor problems and errors.

Table 1: Please change standard error (SE) to standard deviation.

Line 308 : Please check SNP number (6,218,130). It seems to be “62,181 3’ untranslated regions.”

Line 311 - 318 : Fst, Nucleotide diversity parameters were used inappropriately. These are data and evidence that cannot support the author’s argument. If you can’t explain it properly, I suggest removing the content and data.

Line 333 : Please represent the r2 value decreased to half of the maximum value.   

Line 350-353 : Add a space between the QTL name and bracket.

Figure 5-7 : All threshold lines in Manhattan plots should be adjusted. Please draw line based on each p-value of association analyses when the q-values (FDR) are 1 x 10-5.

Figure 7 : Manhattan plot of FLL in Figure 7a doesn’t seem to match Manhattan plot in Figure 5. Please check the plot of FLL.

Result 3.6 and Figure 8 : Because only one variant was detected in LOC_Os05g34660 region, “haplotype” (a group of polymorphisms) is an incorrect term. It should be replaced by “allele”. Please modify all contents (description, figure, and table) related to this gene.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf