A Quick Computational Statistical Pipeline Developed in R Programing Environment for Agronomic Metric Data Analysis

Data harvesting, data pre-treatment and as well data statistical analysis and interpretation are strongly correlated steps in biological and as well agronomical experimental survey. In view to make straightforward the integration of these procedures, rigorous experimental and statistical schemes are required, playing attention to process data typologies. Numerous researchers continue to generate and analyse quantitative and qualitative phenotypical data in their agronomical experimentations. Considering the impressive heterogeneity and as well size of that data, we proposed here a semi-automate analysis procedure based on a computational statistical approach in R programming environment, with the purpose to provide a simple (programmer skills are not requested to users) and efficient (few minute are needed to get output files and/or figures) and as well flexible (authors can add own script and/or bypassed some functions) tool pointing to make straightforward heterogenic metric data interactions in biostatistics survey. The pipeline starts by loading a row data matrix followed by data standardization procedure (if any). Next, data were processed for a multivariate descriptive and as well analytical statistical analysis, comprising data quality control by providing correlation matrix heat-map and as well as p-value clustering analysis graphics and data normality assessment by Shapiro-Wilk normality test. Then, data were handled by principal component analysis (PCA) including PCA n factor survey in discriminating needed factors component explaining data variability. Finally data were submitted to linear and/or multiple linear regression (MLR) survey with the purpose to link mathematically managed data variables. The pipeline exhibits a high performance in term of time saving by processing high amount and heterogenic quantitative data, allowing and/or providing a complete descriptive and analytical statistical framework. In conclusion, we provided a quick and useful semi-automatic computational bio-statistical pipeline in a simple programming language, exempting the researchers to have skills in advanced programming and statistical technics, although it is not exhaustive in terms of features.