Locality-sensitive hashing enables signal classification in high-throughput mass spectrometry raw data at scale

Raw data of nanoLC-IMS-MS/MS (DDA-PASEF) from HeLa whole proteome digest.   

HeLa cells were lysed in a urea-based lysis buffer (7 M urea, 2 M thiourea, 5 mM dithiothreitol (DTT), 2% (w/v) CHAPS) assisted by sonication for 15 min at 4°C in high potency using a Bioruptor instrument (Diagenode). Proteins were digested with Trypsin using a filter-aided sample preparation (FASP) [Wisniewski et al., 2009] as previously detailed [Distler et al., 2016]. 200 ng of peptide digest were analyzed using a nanoElute UPLC coupled to a TimsTOF PRO MS (Bruker). Peptides injected directly in an Aurora 25 cm x 75 µm ID, 1.6 µm C18 column (Ionopticks) and separated using a 120 min. gradient method at 400 nL/min. Phase A consisted on water with 0.1% formic acid and phase B on acetonitrile with 0.1% formic acid. Sample was injected at 2% B, lineally increasing to 20% B at 90 min., 35% B at 105 min., 95% at 115 min. and hold at 95% until 120 min. before re-equilibrating the column at 2%B. The MS was operated in DDA-PASEF mode [Meier et al., 2018], scanning from 100 to 1700 m/z at the MS dimension and 0.60 to 1.60 1/k0 at the IMS dimension with a 100 ms TIMS ramp. Each 1.17 sec MS cycle comprised one MS1 and 10 MS2 PASEF ramps (frames). The source was operated at 1600 V, with dry gas at 3 L/min and 200°C, without nanoBooster gas. The instrument was operated using Compass Hystar version 5.1 and timsControl version 1.1.15 (Bruker).