De novo transcripts were generated by Cufflinks using STAR alignments. The BAM alignments files were filtered using 'samtools rmdup -S' command to remove (PCR?) duplicates. Cufflinks 1.0.3 was used to assemble and quantify transcript models. The transcripts were assembled with the bio-replicas pooled using Cufflinks 1.0.3. The assembled transcripts were then quantified using Cufflinks 1.0.3 separately for each replicate. Bias correction was not used in quantifications. The resulting FPKM1 and FPKM2 values are included as attributes in the GTF file. Contact: dobin@cshl.edu