Informatics View on the Challenges of Identifying Missing Proteins from Shotgun Proteomics

详细信息    查看全文

• 作者：Wai-Kok Choong ; Hui-Yin Chang ; Ching-Tai Chen ; Chia-Feng Tsai ; Wen-Lian Hsu ; Yu-Ju Chen ; Ting-Yi Sung
• 刊名：Journal of Proteome Research
• 出版年：2015
• 出版时间：December 4, 2015
• 年：2015
• 卷：14
• 期：12
• 页码：5396-5407
• 全文大小：521K
• ISSN：1535-3907

文摘

Protein experiment evidence at protein level from mass spectrometry and antibody experiments are essential to characterize the human proteome. neXtProt (2014-09 release) reported 20鈥?55 human proteins, including 16鈥?91 proteins identified at protein level and 3564 proteins unidentified. Excluding 616 proteins at uncertain level, 2948 proteins were regarded as missing proteins. Missing proteins were unidentified partially due to MS limitations and intrinsic properties of proteins, for example, only appearing in specific diseases or tissues. Despite such reasons, it is desirable to explore issues affecting validation of missing proteins from an 鈥渋deal鈥?shotgun analysis of human proteome. We thus performed in silico digestions on the human proteins to generate all in silico fully digested peptides. With these presumed peptides, we investigated the identification of proteins without any unique peptide, the effect of sequence variants on protein identification, difficulties in identifying olfactory receptors, and highly similar proteins. Among all proteins with evidence at transcript level, G protein-coupled receptors and olfactory receptors, based on InterPro classification, were the largest families of proteins and exhibited more frequent variants. To identify missing proteins, the above analyses suggested including sequence variants in protein FASTA for database searching. Furthermore, evidence of unique peptides identified from MS experiments would be crucial for experimentally validating missing proteins.