Ulture. Cell pellets of seed culture had been then harvested by centrifugation (1000 ?g for 10 min) and cultivated in two modified Gg-8 media, i.e. Gg-8 medium containing two.5 M NaCl for the hyper-osmotic strain study and Gg-8 medium containing 0.5 M NaCl for the hypo-osmotic strain study, respectively.Biomass determinationTwo varieties of models had been produced for studying the effects of light good quality, nitrogen availability and osmotic tension on lutein productivity (the response variable). The predictor variables were the percentage of blue light, the level of KNO3 along with the amount of NaCl. The initial model was a conventional quadratic model where the model parameters had been obtained with least squares regression. The second model was a nonparametric model, meaning that no assumptions are made regarding the data generating mechanism. The model was depending on gradient boosted regression trees which have received considerable attention in current years for their superior predictive overall performance and their usefulness in information exploration [36]. The boosted tree model was obtained with all the GBM package for R [37]. The GBM parameters were set as follows: Squared error loss was employed, the amount of trees was 700 (determined by minimizing the out-of-bag error), shrinkage was set to 0.005, the subsampling fraction to 0.5 and three-way interactions had been made use of.Additional fileAdditional file 1: Figure S1 Linear growth of adapted D. salina (HI 001) beneath a total light intensity of 170 E/m2/s red LED light. Figure S2 Correlation between lutein productivity and biomass productivity of D. salina cells (data shown in Table 2 and Table S1). Table S1 Biomass productivity of D. salina in RSM experiments a, Table S2 Strength of variable interactions for the boosted tree model (larger values indicate extra strength), Table S3 Comparisons in between values predicted by the quadratic model and also the experimental data, Table S4 Prediction of maximum lutein productivity by the quadratic model, Table S5 Comparison of optimal circumstances predicted for lutein production by RSM and situations developed for carotenoids production by prior ALE. Quadratic model in coded values (Equation S1).Alga samples of culture suspension had been filtered and collected on a mixed cellulose membrane (pore size: 0.45 m), washed with de-ionized water twice and dried overnight at 60 just before weighing [7].Determination of cell sizeThe cell size was measured by a Countess automated cell counter (Life Technologies Corporation, Carlsbad, California, U.S.A.). D. salina cell size was detected in bead mode devoid of applying trypan blue dye staining.Chlorophyll and carotenoid analysisAbbreviations ALE: Adaptive laboratory evolution; ERK: Extracellular signal-regulated kinase; LED: Light-emitting diode; PBRs: Photobioreactors; ROS: Reactive oxygen species; RSM: Response surface methodology; UPLC-UV-MS: Ultra-performance liquid chromatography, UV and mass spectrometry.6-Bromoquinolin-8-amine web Competing interests The authors declare they have no competing interests.Formula of 1-Acetoxy-1,2-benziodoxol-3-(1H)-one Authors’ contributions WF developed the study, carried out a lot of the experiments and wrote the manuscript.PMID:23453497 GP and MM assisted in evaluation of algal samples with LC/MS. EAS participated in some development experiments. SG participated in modelling operate. SG, OSA and SB contributed to editing and revising the manuscript. B and SB conceived in the study and helped to finalize the manuscript. All authors study and authorized the manuscript. Acknowledgments This study was supported by the Icelandic Technolog.