Input file format for RSEM

[edge]

Hi,

Currently I'm focus on expression profiling of transcriptome study.
As I know, RSEM calculation is based on raw read map back to assembled contig.
My main concern is that whether is advisable that my raw read is in proper pair form and complete length (eg. all same length 100 nt and present in pair form)?
Since it is always a good practice to pre-process the data-set (eg. trim adaptor sequence and low quality bases) before running through de-novo assembly.

Unfortunately, trim adaptor might cause the raw data in different read length and remove low quality read will cause that those so-called "high quality" read is not in proper pair form (some will only keep left read and some will only keep right read).

Thus I just not sure whether the best practice is I just keep those high quality read in proper pair form and complete length for running de-novo assembly and RSEM.
Since shorter read length (due to trim adaptor sequence) is more easier to map back to contig.
I just scare the biased of data mapping cause the expression result inaccurate.

Thanks for any advice sharing regarding RSEM program utilization.

[edge]

I got two sets of Illumina pair-end transcriptome data (both set are closely-related plant species). First, I will trim adaptor sequence and low quality bases of both data set separately before de-novo assembly separately.
I got two sets of assembly contigs and two sets of Illumina pair-end data right now.
My main objective of research is to identify the gene profiling of two closely-related plant .
I will use RSEM output file as an input file for EBSeq (as mentioned by RSEM) in order to run differential expression.

Thanks for any advice.