Recently, liquid chromatography coupled to mass spectrometry (LC-MS) has become a standard technique for identifying differential abundance of peaks as biomarkers. Two major problems in the preprocessing of LC-MS data analysis are how to adjust and align multiple LC-MS datasets efficiently and correctly. Hence, an effective algorithm is needed to adjust the variation in retention time and align protein signals automatically. In this study, we proposed a novel algorithm, PeakAlign, based on a clustering technique for adjusting the shifted peaks and aligning the same protein signals from different samples. The PeakAlign algorithm consists of two phases, namely adjustment phase and alignment phase. In the adjustment phase, a LOESS regression method is used to adjust the shifting trend among peaks. In the alignment phase, a cluster-based technique is applied to align the adjusted peaks. For experimental evaluation, two different alignment approaches, SlidingWin algorithm and DTW algorithm, were implemented. Through analyzing the real LC-MS dataset, we demonstrate the usefulness of our proposed algorithm, PeakAlign, on the LC-MS-based samples.