What if I split the BAM file by chromosome using following command samtools view -b input.bam chr1 chr2 chr3 chr4 > output.bam and then Cufflinks is done. Would it produce false positive? Can somebody help me understand or give me some reference article? Actually it has been hampering my work for long and I would be extremely grateful.
As far as I know Cufflinks is in a deprecated state and you should strongly consider the alternative of Stringtie - Better performance overall. Cufflinks works on RABT assembly so splitting by chromosome shouldn't cause too many significant issues. Your FP rate will depend more on your method of differential expression. Using Stringtie, you can prepare your abundance estimations for analysis with DESeq2, which is highly reputable in analysing RNA Seq data.