Candidate Gene Analysis
By taking into consideration the results of the GO, MapMan and PPI network, stress responsive DEGs having function relevant to the cell wall, plant-type vacuole membrane, water transport, antioxidant activity, auxin transport, peroxidase activity, protein kinase activity, TF activity, secondary metabolite synthesis, hormone signaling and signaling were screened (Table 2 ). A total number of 981 DEGs associated with above-mentioned functions were screened, and used for candidate gene prediction analysis. Gene ID, annotation and expression pattern of 981 genes in all four genotypes viz., DTP, DSP, DTL & DSL are presented (Table 2; Table S7-S13) . These all 981 DEGs were subjected to cluster analysis by using dChiP software, and were grouped into six different clusters viz., I, II, III, IV, V & VI on the basis of expression pattern of DEGs in four different genotypes (Fig. 9). Out of these six clusters, the DEGs in Cluster II & III showed consistent opposite expression trend between tolerant (DTP & DTL) and sensitive (DSP & DSL) genotypes. For example, the DEGs of cluster II revealed up-regulation in tolerant genotypes (DTP & DTL) and down-regulation in sensitive genotypes (DSP & DSL), whereas the DEGs of cluster III showed down-regulation in tolerant genotypes but up-regulation in sensitive genotypes. The cluster II & III include 547 DEGs out of total 981 DEGs, and these 547 genes showed consistent and opposite expression pattern in tolerant and sensitive genotypes (Fig. 9; Table 4 ), suggesting vital role of these genes in the regulation of drought tolerance in soybean.
To further predict the most possible genes among above 547 DEGs, the Chr.08 earlier reported to be rich in QTLs and “Hotspot regions” associated with drought stress tolerance at different growth stages in soybean such as germination and early seedling (V1) were screened for these DEGs (Wang et al ., 2012; Manavalan et al ., 2015; Valliyodan et al ., 2016). By taking this into consideration, we screened all the 547 DEGs, and 28 of them were found on Chr.08(Table S14). These 28 genes were further subjected for sequence polymorphism (such as SNP, InDels) detection between drought-tolerant (DTL) and drought-sensitive (DSL) lines. Out of these 28 genes, only eight genes showed non-synonymous SNPs viz., Glyma08G162700 ,Glyma08G091400, Glyma08G097300, Glyma08G045000, Glyma08G271600 ,Glyma08G014200 , Glyma.08G201700 and Glyma08G042100 . Among these eight genes, two genes viz., Glyma08G091400 andGlyma08G097300 were located within “Hotspot region” on Chr.08 related with drought tolerance in soybean (Manavalan et al . 2015; Valliyodan et al . 2016). The function annotation of eight plus two hub genes (Glyma.08G029400 /MYB127 andGlyma.10G184500 /MYB78) are presented in Table 5.
Hence, based on the PPI, GO, dChiP, gene annotation and sequencing analysis, these above ten genes were predicted as candidate genes regulating drought tolerance at germination stage in soybean (Table 5 ). Therefore, further validation of these genes is required before they can be successfully employed for molecular breeding of drought tolerance in soybean.