Look for references from Krokhin and Spicer, who developed a framework for predicting relative (not absolute!) retention times of peptides in RPLC under different conditions of stationary and liquid phase, and an algorithm called SSRcalc (http://hs2.proteome.ca/SSRCalc/SSRCalcX.html) to start with. Similar concepts are implemented in e.g. iRT, however you usually need to calibrate your retention time scale using peptide standards.

I'm fully agree with Christof. In my experience, predict retention time of any peptide in a matrix, on a particular system, is very difficult. Best option that i found is to calibrate your system with iRT (using Skyline, http://goo.gl/zHYEoM). If you don't want to use an iRT standards kit, maybe you can calibrate with abundant peptides in your matrix. I successfully used this strategy to predict retention time of low concentrate peptides, with analysis of abundant peptides to calibrate my iRT.