2.5.2 Analysis of RNA‐Seq data
The raw RNA‐Seq reads were processed to remove sequences with adapter contamination, low‐quality nucleotides in excess of 10% and unknown nucleotides greater than 50 using Trimmomatic (Bolger et al., 2014). Processed sequences with length less than 40 bp were also got rid of. Ribosomal RNA sequences were removed after been discovered by aligning reads to ribosomal RNA database (Quast et al., 2012). The remaining clean reads were mapped to reference tomato genome using HISAT software that allowed up to two mismatches (Kim et al., 2015). The expression level of each gene was determined by counting the number of fragments that mapped to each gene and then normalized to number of fragments per kilobase of transcript sequence per millions (FPKM) base pairs sequenced using HTSeq software. A gene with FPKM value ≥ 0.1 was considered expressed. To recognize differentially expressed genes (DEGs), the FPKM values of each gene from WW and DS anthers were analyzed using DESeq software (Anders and Huber, 2010). A rigid cut‐off, ǀ log2 fold change ǀ >1 and p-adjusted < 0.05 was set as thresholds to consider a gene significantly differentially expressed. GOseq (Young et al., 2010) was used to analyze functional enrichment of specific gene ontology (GO) terms for DEGs. KEGG pathways significantly enriched with DEGs were determined using the KOBAS software.