A comprehensive CHO SWATH-MS spectral library for robust quantitative profiling of 10,000 proteins

Sequential window acquisition of all theoretical fragment-ion spectra (SWATH) is a data-independent acquisition (DIA) strategy that requires a specific spectral library to generate unbiased and consistent quantitative data matrices of all peptides. SWATH-MS is a promising approach for in-depth proteomic profiling of Chinese hamster Ovary (CHO) cell lines, improving mechanistic understanding of process optimization, and real-time monitoring of process parameters in biologics R&D and manufacturing. However, no spectral library for CHO cells is publicly available. Here we present a comprehensive CHO global spectral library to measure the abundance of more than 10,000 proteins consisting of 199,102 identified peptides from a CHO-K1 cell proteome. The robustness, accuracy and consistency of the spectral library were validated for high confidence in protein identification and reproducible quantification in different CHO-derived cell lines, instrumental setups and downstream processing samples. The availability of a comprehensive SWATH CHO global spectral library will facilitate detailed characterization of upstream and downstream processes, as well as quality by design (QbD) in biomanufacturing. The data have been deposited to ProteomeXchange (PXD016047).

High pH reverse phase liquid chromatography (bRPLC) peptide fractionation. A total of 200 µg of peptide samples from CHO-K1 WCL and subcellular-organelle fractions were each subjected to a 60-min multi-step gradient performed on a Kinetex core shell C18 column (2.6 µm, EVO C18, 150 mm × 3.0 mm, 100 Å, Phenomenex, Brechbuhler AG, Schlieren, Switzerland) using an UltiMate 3000 UHPLC system (Thermo Fisher Scientific). The mobile phase was 10 mM ammonium formate (buffer A, adjusted to pH 10 using ammonium hydroxide) and 10 mM ammonium formate in 80% acetonitrile (ACN) (buffer B, adjusted to pH 10 using ammonium hydroxide). The peptide samples were separated with a gradient from 0-35% buffer B for 55 min at a flow rate of 0.8 mL/min. The fractions were collected and pooled in concatenation into 30 fractions for CHO-K1 WCL; and 10 fractions for each subcellular-organelle sample. The fractions were dried in a SpeedVac vacuum concentrator at room temperature and stored at −80 °C for future use.   Fig. 1 Workflow for creating and using the SWATH CHO global spectral library. The CHO-derived samples were processed using in-house multi-dimensional separation protocol. Briefly, the CHO-K1 cells were lysed and fractionated using differential ultracentrifugation to isolate nuclear (NE), mitochondrial (MITO), and heavy-membrane (HM) compartments. The protein lysates from whole cell (WCL) and subcellularorganelle compartments were tryptic digested, subsequently fractionated using basic reverse-phase liquid chromatography separation, and subjected to DDA-MS analysis. Protein digest from harvested cell culture fluid (HCCF) and downstream processing (DSP) mAb samples were directly subject to SWATH-MS in TripleTOF 6600. The raw DDA data was searched locally in ProteinPilot TM software and the results were uploaded to OneOmics TM for spectral library construction. The SWATH-MS data sets were processed locally using PeakView ® and MarkerView TM or using OneOmics TM . The applicability and robustness of the CHO global spectral library were evaluated with SWATH-MS data sets of different CHO-derived samples, including WCL of different cell lines, HCCF and DSP mAb samples, and using various LC-MS instrumental setups.
www.nature.com/scientificdata www.nature.com/scientificdata/ NanoSpray III Ion Source (SCIEX) and interfaced with a Waters nanoAcquity UPLC system (Waters, Milford, MA) or Eksigent Ekspert nanoLC 425 (NanoLC Ultra 2D, Eksigent, Toronto, Canada). The peptide samples were separated at 50 °C on an ACQUITY UPLC M-class peptide BEH C18 column (75 µm x 200 mm, particle size of 1.7 µm, and pore size of 130 Å). The iRT reference peptides (Biognosys AG, Schlieren, Switzerland) were added to all samples at 1:10 ratio prior to MS injection for retention time calibration. The LC system was operated with 0.1% formic acid (FA) in water (buffer A) and 0.1% FA in ACN (buffer B) at a flow rate of 300 nL/min. The separation gradient was from 5-35% of buffer B over the period of 110 min. The MS was operated in DDA top 20 mode with the following parameters: MS1 spectra were collected at 400-1,500 m/z for 500 ms, 20 most intense precursors with charge states 2-5 that exceeded 125 counts/s were selected for fragmentation, and the corresponding fragmentation MS2 spectra were collected at 50-2,000 m/z for 100 ms. Rolling collision energy (equation: (0).0625 × m/z -10.5) (derived from SCIEX) with a collision energy spread (CES) of 15 eV was set as the fragmentation patterns used in SWATH-MS analysis.
Data-independent acquisition (DIA) SWATH-MS. SWATH-MS acquisition was operated using the same LC-MS instrumental setup as described above with some modifications. Briefly, a 100-variable-window setup was generated using the SWATH ® Variable Window Calculator 1.1 (SCIEX) with a 1 m/z window overlap on the lower side of the window. The MS1 survey scan was acquired from 400-1,250 m/z for 250 ms and MS2 spectra were acquired in high-sensitivity mode from 100-1,500 m/z for 30 ms. The total cycle time was ~3.3 s. The collision energy used in SWATH-MS acquisition was that applied to a doubly charged precursor centered in the middle of the isolation window calculated with the same collision energy equation for DDA, and with a CES of 5 eV. For the analyses conducted in capillary flow and microflow rate, the SWATH-MS data were recorded on a TripleTOF 6600 mass spectrometer coupled to a DuoSpray Analytical Ion Source (SCIEX). For capillary flow LC setup, the Eksigent Ekspert nanoLC 425 system was connected to a Waters CSH C18 column (300 µm × 150 mm, 1.7 µm, 130 Å) or a Eksigent ChromXP C18 column (300 µm × 150 mm, 3 µm, 120 Å) and operated in trap-elute mode for 1-h gradient at a flow rate of 5 µL/min. For the microflow LC setup, a Waters nanoAcquity LC system connected with a Waters CSH C18 column (1 mm × 150 mm, 1.7 µm, 130 Å) was utilized and the system was operated in direct-inject mode at a flow rate of 50 µL/min. The 100-variable-windows setup was applied and optimized in both the capillary flow and microflow rate instrumental system.
Generation of CHO spectral library. The CHO spectral library was constructed using the workflow established by SCIEX 32 . Briefly, the raw DDA data files were processed using the ProteinPilot TM software (version 5.0.1) against a Chinese hamster (CH) protein sequence database. The database was the latest release of CH RefSeq assembly (downloaded on August 2019 from ncbi.nlm.nih.gov; GCF_003668045.1_CriGri-PICR_protein.faa), appended with the Biognosys iRT fusion protein sequence and in-house mAb heavy-and light-chain protein sequences. The alkylation reagent was iodoacetamide. The search effort setting was Rapid ID with carbamidomethyl (C) as a fixed modification, and oxidation (M), deamidation (NQ) and pyroglutamic acid conversion (EQ) as variable modifications. Each of the DDA raw data files was searched in ProteinPilot TM and the result files were used as the input libraries for spectral library generation. The input libraries and SWATH-MS acquisition data were uploaded to the Illumina BaseSpace Cloud (www.basespace.illumina.com) using the CloudConnect software and processed in SCIEX Cloud OneOmics TM (Fig. 1). For the construction of spectral library, each input library was filtered by 1% FDR and 99% confidence threshold to remove low-confidence peptide identifications. The largest filtered input library, which contained the highest number of high-confidence peptides, were selected as base, and peptides identified from smaller filtered input libraries were merged to the base library using a non-linear calibration strategy in the OneOmics TM 32 . Any newly identified peptides are added to the existing proteins and any newly detected proteins are added if not present in the base library. The merged spectral library was further processed with an in-house script to identify and remove any shared and/or modified peptides.
The combined spectral library was constructed by searching all the 63 DDA raw data files together in the ProteinPilot TM software. The 1D spectral library was generated using unfractionated peptide mixtures from a CHO-K1 WCL sample.

SWatH-MS data analysis.
The SWATH-MS data analysis was performed in the OneOmics TM or the SWATH TM Processing software in local PC. In the OneOmics TM , the CHO endogenous peptides were extracted according to the precursor m/z, intensity and confidence of identification across the entire time range, and the best scoring peak groups were used for RT calibration. The data was filtered by 1% FDR and the comparisons of proteins and/or peptides were further filtered by 20% coefficient of variation (CV) cutoff between replicates. In local SWATH-MS processing workflow, the spectral library and SWATH-MS acquisition data were loaded into SWATH TM Processing microApp in PeakView ® software. The iRT, mAb and shortlisted endogenous CHO peptides were manually selected as reference peptides for RT calibration. The peak groups were extracted with a 99% peptide confidence threshold and a 1% peptide FDR cutoff. The RT extraction window and the fragment ion mass tolerance were set to 5 min and 75 ppm, respectively. After data extraction, the results were imported into MarkerView TM (version 1.3.1) for further data processing and analysis. Microsoft ® Excel, Python programming and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis 33 were also applied in the downstream data browsing and analysis.

Data Records
The raw DDA data files for library generation, the search result (group files), the CHO spectral library, and the SWATH-MS acquisition data applied in the current study have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) 34 via the PRIDE partner repository 35,36 with the data set identifier PXD016047 28 . (2020) 7:263 | https://doi.org/10.1038/s41597-020-00594-z www.nature.com/scientificdata www.nature.com/scientificdata/ technical Validation error rate control during spectral library generation. The number of proteins and peptides identified from the raw DDA data files are significantly affected by the quality and redundancy of the protein sequence database used. UniProtKB/Swiss-Prot 37 , a freely accessible database of protein sequence and functional information, is manually annotated and reviewed to contain high-quality and non-redundant protein information and hence highly recommended for the purpose of public spectral library generation. However, the UniProtKB/Swiss-Prot contains only 14 and 237 reviewed entries for Chinese hamster/CHO-K1 and Cricetulus griseus, respectively. We therefore used the most updated NCBI reference sequence (RefSeq) proteome database of Chinese hamster in this study. NCBI RefSeq 38 is also a curated non-redundant collection of sequences which records genomes, transcripts and proteins information from multiple sources. We searched the mass spectra from the DDA raw data against the Chinese hamster proteome database containing 46,750 protein entries. The error rate in the analysis of these large sample cohorts was carefully handled to prevent the accumulation of false positive identifications during the construction of the spectral library. We utilized the library merging algorithm established in the OneOmics TM to filter the lists of peptides in ProteinPilot TM group files to a 1% global protein FDR 32 . All the peptides that passed the 99% peptide confidence threshold were selected and merged to the base library using a non-linear retention time calibration strategy 32 . Any shared or modified peptides (except those with carbamidomethylation on cysteine residue) were identified and removed from the spectral library using an in-house Python script to avoid any possible redundancy and inaccuracy in SWATH-MS data extraction and quantification using the spectral library. A total of 10,974 proteins, comprised of 199,102 peptides, 247,135 precursors and 741,405 transitions, were represented in the finalized library, which was known as CHO global spectral library hereafter.
Coverage of the CHO global spectral library. During the construction of CHO global spectral library, it was found that the number of confidently quantified proteins (at 20% CV cutoff and 1% peptide FDR) reached plateau despite introducing more DDA files into the spectral library (Fig. 2a). As shown in Fig. 2a, the Subcell library (contained 30 fractions of bRPLC-separated subcellular compartments and 1D library) quantitated 9.18% more proteins than the CL library (contained 30 fractions of bRPLC-separated cell lysate and 1D library). By using the CHO global spectral library (consisted of the fractions from Subcell library, CL library and 1D library), we quantitated slightly more proteins (8.34%) than the previous work (Fig. 2a). It is noteworthy that a large number of library inputs will critically increase the time required for loading libraries into OneOmics TM . Therefore, www.nature.com/scientificdata www.nature.com/scientificdata/ we did not further fractionate the CHO-K1 protein lysate as these DDA inputs have provided sufficient CHO proteome coverage while allowing reasonable downstream SWATH-MS data processing time. In summary, the CHO global spectral library was constructed by using the MS/MS spectral data obtained from the unfractionated (1D) CHO-K1 cell lysates as the base, followed by adding the MS/MS spectral data sets derived from the three subcellular organelles (10 fractions of nuclear compartment, 10 fractions of mitochondrial compartment and 10 fractions of heavy-membrane compartments) and an extensively separated CHO-K1 whole-cell protein lysate (30 fractions).
The CHO global spectral library has reached a 23.48% proteome coverage of the 46,750 protein entries represented in the CHO RefSeq proteome database. This is 2.5 times more proteins and 8.8 times more peptides than a non-fractionated 1D spectral library (comprising of 3,134 proteins and 20,249 peptides identification) ( Table 1). We also compared the CHO global spectral library (which was built based on the CHO-K1 cell proteome) to the latest proteomics study of CHO-DG44 and CHO-S cell lines (Table 1) 13 . When filtered with at least two unique peptides, the CHO global spectral library identified significantly more proteins (+56.82% to +72.42%) than individual CHO-DG44 and CHO-S data set, respectively, and a slightly higher number of proteins (+2.03%) than the combination of these two data sets. Remarkably, the CHO global spectral library represented a 242.52% to 313.03% improvement over the previous work in peptide identification ( Table 1). The CHO global spectral library is thus able to provide a significantly higher number of empirically detected peptide identities for better protein identification and quantification. As displayed in Fig. 2b, the CHO global spectral library showed high similarity in the distribution of the number of assays per protein as that of 1D library, but the former covered a wider effective separation gradient with its large number of peptide identities (Fig. 2c).
A CHO-K1 WCL SWATH-MS data set was processed with the CHO global spectra library. Up to 3,478 proteins (65.86%) could be confidently quantified (at 20% CV cutoff and 1% peptide FDR) with at least two unique peptides (Fig. 2d). Additionally, we perceived the paramount significance to quantitatively identify the critical components in CHO biological pathways relating to the stability, quality and productivity of recombinant protein production as well as the cell viability. By using KEGG pathway mapping analysis, we demonstrated that proteins associated with Protein Processing in ER (83.81%, Fig. 3a), Cell Cycle (82.98%, Fig. 3b) and N-glycan Biosynthesis (82.05%, data not shown) were well represented in the CHO global spectral library (proteins boxed in yellow color in Fig. 3). In addition, more than half of these proteins could be confidently identified (at 20% CV cutoff and 1% peptide FDR) in the SWATH-MS analysis (proteins highlighted in red color in Fig. 3).

Calibrated retention time in CHO global spectral library.
One of the advantages of adopting nanoflow instrumental setup in the construction of SWATH spectral library was that nanoflow LC-MS provides the highest sensitivity in DDA-MS without requiring huge amounts of starting materials as compared to higher flow LC-MS. However, retention time drift 39 is an inherent issue in the nanoflow instrumental setup which would lead to retention time misalignment in combined database search (the first step in constructing spectral library). Therefore, we utilized the library merging algorithm in OneOmics TM to improve the retention time correlation and alignment between input libraries during the library construction. In subsequent SWATH-MS data extraction procedure in OneOmics TM , a non-linear autoRT calibration algorithm was applied 32 . Briefly, one hundred high-abundance, confidently detected peptides, including those of iRT peptides, mAb protein and endogenous CHO proteins, were selected as reference peptides across the entire scan time range. The retention time frame of spectral library was matched to that of SWATH-MS data based on the reference peptides to fine-tune the retention time from run to run. Here we compared the CHO global spectral library and the uncalibrated combined-searched spectral library in the SWATH-MS analysis of CHO-K1 WCL data set. The CHO global spectral library identified a total of 6,445 proteins, which was 49.54% more than that of the combined-searched library (1% peptide FDR) ( Table 2). At 20% CV cutoff, the CHO global spectral library quantitated higher numbers of protein (4,717) and peptide (16,919) identities as compared to those of the combined-searched spectral library (2,537 proteins and 8,252 peptides) ( Table 2). On top of that, the former has reported lower median CV values in both protein-and peptide-level quantitation ( www.nature.com/scientificdata www.nature.com/scientificdata/ merging and data extraction procedure. By using the CHO global spectral library which contained highly accurate predicted retention times, we could execute correct and consistent SWATH-MS data analysis. Portability of the CHO global spectral library. Different laboratories may acquire their SWATH-MS data sets using various instrumental setups. For example, in biopharmaceutical industries, higher flow-rate LC-MS analytical instruments are often preferred in order to achieve high reproducibility and throughput with  www.nature.com/scientificdata www.nature.com/scientificdata/ a shorter duration of data acquisition and processing. Since the CHO global spectral library was established in nanoflow LC-MS setup (TripleTOF 6600 coupled to Waters nanoAcquity LC), we validated its performance with three CHO-K1 WCL data sets acquired in nanoflow LC (Eksigent) coupled to TripleTOF 6600 (NF6600 data), capillary flow LC (Eksigent) coupled to TripleTOF 5600+ (CF5600 data), and microflow LC (Waters) coupled to TripleTOF 6600 (MF6600 data) ( Table 3). Noted that in CF5600 and MF6600 we analyzed higher loading amounts of sample to compensate for the lower sensitivity at higher flow rate LC-MS. The relationship between the delta retention times (∆RT = observed retention times -predicted retention times) and the predicted retention times were studied. As shown in Fig. 4a, the ∆RT maintained within ±2.5 minutes across the entire predicted retention time range (red colored "x" indicated the eleven iRT reference peptides). Graphical demonstration using violin plot showed the distribution of observed and predicted retention times were highly consistent in the three LC-MS setups (Fig. 4b). Total 4,222 proteins were confidently quantified from three data sets (at 20% CV cutoff and 1% peptide FDR): In NF6600 there were 3,631 proteins (15,311 peptides) quantified when 1 µg of WCL was analyzed; In CF5600 3,519 proteins (12,825 peptides) were quantifiable from 5 µg of WCL; In MF6600 up to 4,087 proteins (17,071 peptides) were quantified from 20 µg of WCL (Fig. 4c). Hundreds to thousands of proteins and peptides could be consistently quantified in three data sets when more stringent CV cutoffs (5 and 10%) were applied (Fig. 4c). Right-skewed distribution of CV values with low median CVs ranging from 9.5% to 12.3% in both the protein-and peptide-level quantification demonstrated the portability as well as the robustness of the CHO global spectral library for the analyses of SWATH-MS data sets acquired across three LC-MS setups (Fig. 4d). These results have further elaborated the effectiveness of retention time calibration in our pipeline, and supported the potential transfer of the CHO global spectral library to industrial scale application which often requires high throughput with short sample and data handling duration.
In addition to CHO-K1 cells, there are other CHO-derived cell lines that are commonly studied and widely utilized in the biomanufacturing process. We evaluated the performance of the CHO global spectral library, which was built using the CHO-K1 cell proteome searched against the Chinese hamster RefSeq database, in the SWATH-MS analysis of another two CHO cell lines, namely CHO-DG44 and CHO-S. In this analysis, more than 4,700 proteins were confidently quantified (at 20% CV cutoff and 1% peptide FDR). A total of 1,664 proteins (35.24%) were commonly identified in the three cell lines and 2,979 proteins (63.09%) were quantified in at least two cell lines (Fig. 4e). The median CV values for the protein abundances in each cell line ranged from 8.72% to 12.28% while the median CV values for all the quantified proteins across the three cell lines was 9.97% (Fig. 4f). The SWATH-MS results of CHO-DG44 and CHO-S cells suggested that the CHO global spectral library can be effectively used to quantify thousands of proteins across multiple CHO cell lines.

Robustness of protein identification and quantification in CHO HCCF and HCPs in DSP mAb
samples. Being the most widely utilized host cell in biopharma mAb production, the systematic and  www.nature.com/scientificdata www.nature.com/scientificdata/ www.nature.com/scientificdata www.nature.com/scientificdata/ comprehensive analysis of the intracellular CHO proteome using DIA SWATH-MS will help to better understand this model cell line and contribute towards the rational improvement of CHO cells performance 40 . In addition, the profiling of HCCF protein dynamics will also facilitate the bioprocess design and optimization, and the monitoring of HCP impurity in the biologics product 41   www.nature.com/scientificdata www.nature.com/scientificdata/ over a period of four months. A total of 360 SWATH-MS runs were carried out in six batches of experimental groups, E01 -E03 were separated using Waters CSH C18 column; while E04 -E06 were separated using Eksigent ChromXP C18 column. Each experimental group consisted of six biological replicates (B01 -B06), and each biological replicate comprised of ten technical replicates (T01 -T10). Except for nine technical replicate runs (four from E05-B06 and five from E06-B06) which were excluded due to spectrum quality failure, total of 351 SWATH-MS data sets were analyzed in OneOmics TM using the abovementioned workflow, and total 4,985 proteins were identified (at 1% peptide FDR) after SWATH-MS data extraction. The CV values for each of the six biological replicates in E01 -E06 were also determined. The violin plot showed highly similar pattern of CV distributions (median CV values of ~35%) although the MS data was acquired over a long period of four months, and two analytical columns from different vendors were used (Fig. 5a). Higher median CVs observed in this experiment might be due to the inherently higher margin of error for quantifying low abundance proteins 42 and much wider dynamic range of protein abundances in the HCCF samples as compared to that of cell lysate samples. We have also observed that lower median CV values (~22%) is achievable when subsets of the replicates were acquired within a shorter duration (data not shown). A Pearson correlation matrix was constructed using the intensities of all identified proteins. While E01 -E03 (180 runs) and E04 -E06 (171 runs) were visually distinct from each other, corresponding to the use of two different analytical columns, the Pearson correlation matrix showed that the 351 runs were positively correlated with each other with a median correlation coefficient r value of 0.88 (indicated as a red colored dash line on the color bar in Fig. 5b and Supplementary Table 1). These results have demonstrated that extracellular CHO proteins can be identified and quantified when performing the SWATH-MS analyses of HCCF using the CHO global spectral library in OneOmics TM .
We next evaluated the performance of CHO global spectral library in monitoring HCP clearance in the DSP workflow. The DSP mAb samples collected from a generic DSP practice, including original material (OM), post protein A eluate (PrA), post cation exchange flow through (CEX) and post anion exchange flow through (AEX), were digested and analyzed accordingly (Fig. 1). ELISA analysis of these samples showed that the total HCP concentrations were greatly reduced from 510,762 ppm in OM, to 1,081 ppm in PrA, 557 ppm in CEX, and finally only 7 ppm in the AEX. In the SWATH-MS analysis, total 1,900 proteins were quantified (at 20% CV cutoff and 1% peptide FDR in OM) from one microgram of each DSP sample, including previously reported difficult-to-remove and problematic HCPs 43-45 (Online-only Table 1). Heat map analysis of all the identified proteins showed that majority of the impurities were removed after protein A affinity chromatography (Fig. 5c). Although some HCPs were found to be enriched together with mAb (at the top and bottom of heat map of PrA), these impurities were properly removed by the following polishing steps to generate a highly purified sample (AEX) (Fig. 5c). The protein concentration (in ppm) of four well-known difficult-to-remove HCPs, including clusterin (Clu), cathepsin D (Ctsd), heat shock cognate 71 kDa protein (Hspa8) and phospholipase B-like 2 (Plbl2), were illustrated as examples: these HCPs were successfully removed after the DSP purification steps (Fig. 5d). The SWATH-MS results were not only in line with the ELISA result, but also provided critical information to monitor individual HCP abundance in each step of DSP purification. Taken all together, the CHO global spectral library can be effectively applied in the identification and quantification of thousands of proteins from CHO-derived samples in our SWATH-MS analysis pipeline.

Usage Notes
Generating alternative SWATH-MS spectral library. The current CHO spectral library has been constructed using SCIEX spectral library generation pipeline and could be directly applied in SCIEX SWATH TM Processing software and OneOmics TM . With the library merging strategy implemented in the OneOmics TM , users could easily expand the proteome coverage of spectral libraries in future. However, future efforts are necessary to expand the library to include cell line-specific protein or peptide identification using other CHO cell lines.
Application of the CHO spectral library in other LC-MS instruments. The CHO spectral library has been successfully applied in the analyses of SWATH-MS data sets acquired using different LC-MS instruments (TripleTOF 5600+) from the same vendor. To apply the spectral library in the DIA analysis using a LC-MS instrument provided by a different vendor, a user needs to ensure the similarity of the ion fragmentation and the optimization of collision energy settings when setting up a DIA acquisition method. Besides, it is recommended to use an analytical column of similar property and spike iRT peptides into the samples for accurate RT alignment and calibration in subsequent data processing.
Absolute protein quantification of HCPs. It is feasible to perform absolute HCP quantification using the CHO global spectral library in the SWATH-MS analysis pipeline. Internal standard proteins with known concentration relative to the mAb product are spiked into the samples prior to LC-MS injection. During data analysis, the CHO global spectral library, appended with the assays of standard proteins, will be used in OneOmics TM . The absolute protein concentration (ppm) of targeted HCPs can then be obtained by directly comparing the protein abundance of HCPs to a calibration curve constructed from the standard proteins.