脂質辨識加速工具：自動處理和辨識液相層析串聯質譜脂體學資料

Abstract

脂質辨識在釐清脂質於疾病和細胞信號傳導中的作用是非常重要的，其中使用的平台又以高通量液相層析串聯質譜為多數。液相層析電噴灑游離串聯質譜（LC-ESI MS / MS）為其中的一種儀器，使用其進行脂質辨識的工作流程包括從特定類別脂質掃描模式(class-specific scan mode)如前驅離子掃描(precursor ion scan)或中性丟失離子掃描(neutral loss scan)中挑選出目標物的荷質比(mass-to-charge ratio, m/z)和滯留時間(retention time, rt)、進行單一產物離子掃描(product ion scan)，並從該結果辨識此質譜為何種脂質所組成。如有多個目標物，就要重復多次產物離子掃描的實驗。上述流程極度仰賴人工，且處理分析這些資料十分耗時、動作重複並需要一些背景知識。在這項研究中，我們開發了一網站工具—脂質辨識加速器（LIA），以加速使用液相層析電噴灑游離串聯質譜進行脂質辨識的過程。 LIA會針對不同掃描模式質譜的資料進行特定的處理。如拿到特定類別脂質掃描模式質譜會先實作資料前處理過程如基線校正(baseline correction)，平滑(smoothing)，去同位素(de-isotope)和峰值檢測，以挑出目標物的荷質比和滯留時間，再規劃多個目標物在產物離子掃描實驗中打碎的順序以實現多目標物的產物離子掃描。如拿到產物離子脂質掃描模式質譜會整合質譜前驅離子資訊和其碎片模式得出多個目標物的脂質辨識結果。使用LIA的線上分析服務可加速脂質辨識的流程，得到具再現性且可信賴的結果，且所有流程所使用的參數都會記錄在資料庫供使用者查詢，LIA的線上分析服務的網址為http://lia.web.cmdm.tw。

Lipid identification based on high-throughput liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI MS/MS) platform is a critical tool in the clarification of the role of lipid in diseases and cell signaling. A typical workflow of LC-ESI MS/MS lipid identification involves four major steps: (1) conducting the lipid class-specific scan (precursor ion scan or neutral loss scan), (2) selecting peaks, which contain the information of the mass over charge (m/z) and retention time, from the spectrum of the lipid class-specific scan modes, (3) generating the candidate lipids list to conduct the product ion scan, and (4) identifying lipid with side chain(s) information from fragmentation spectrum in the product ion scan. However, the process is time-consuming and requires experts’ training and labors at repeating product ion scanning using the candidate lipids list selected from precursor ion scan. In this study, we developed an automatic analysis tool, termed lipid identification accelerator (LIA), to accelerate the labor-intensive process for LC-ESI MS/MS lipid identification by reducing the number of candidate lipids list for product ion scan. A user only needs to upload the results from the lipid class-specific scan. For the spectrum in the lipid class-specific scan, LIA implements baseline correction, smoothing, and peak detection to obtain the m/z and retention time of interests, followed by filtering and de-isotoping to focus on class-specific and monoisotopic m/z with optional matching the user-defined lipid library to the known lipids. It is then proceeded to automatically reduce the number of candidate lipids list and generate the fragmentation order list for further product ion scan. Thus, LIA can help to arrange one single product ion scan with multiple m/z from precursor ion scan more efficiently. If users were targeting at lipid with specific molecular weight, LIA can also automatically identify the lipid by interpreting fragmentation patterns to acquire the side chain(s) information with the product ion scan information provided by the users. With the assistance of LIA, the process time of lipid identification can be greatly shortened. LIA is freely available at http://lia.web.cmdm.tw.