Write a comment

This is an effort based on our monthly Journal Club preprint review with a collected comments/thoughts/suggestions from the group members

General Impressions

The manuscript is concise, reads well, and convincingly motivates the need for SCP on heterogeneous biofluids. The authors clearly invested considerable effort in both sample preparation and data analysis, leveraging state‑of‑the‑art instrumentation.

·      Title vs. data scope

The authors claim to “characterize” heterogeneous cell populations, yet analyzed just 110 saliva and 149 tear cells pooled from ten donors. Because samples are pooled, inter‑donor variation, -arguably the most interesting source of heterogeneity—is lost. if pooling is unavoidable, discuss the resulting limitations explicitly.

Excluding over half of the isolated saliva cells—especially smaller ones—may bias the dataset toward more abundant cell types. This could overlook rare but relevant populations, thereby limiting the method’s ability to fully characterize biofluid heterogeneity. Include an analysis of the excluded saliva cells (size distribution, viability, protein IDs) to show whether important sub-populations were lost.

The Abstract should emphasize whether the optimized workflow uncovered biologically meaningful cell types or pathways. Consider replacing one of the benchmark sentences with a brief statement of the most salient biological insight. This will better align the Abstract with readers’ expectations and the authors’ ambitions.

·      Depth of coverage is eye‑catching, but..

A parallel search with an independent engine (e.g., DIA‑NN) would help convince readers these counts are not inflated. What is the reasoning behind benchmarking with 50 pg of HeLa while most SCP studies benchmark with 200 -250 pg of HeLa.

·      Liquid‑biopsy angle is genuinely exciting

Isolating exfoliated cells from saliva and tears offers a low‑invasive, repeatable sampling route with clinical potential—an elegant way to push SCP beyond cultured lines. Novelty of the study could be highlighted. If this is the first SCP study on these two biofluids, the authors should emphasize that novelty; conversely, they should temper claims until disease‑relevant biology is demonstrated.

·      Overall experimental design

Quite well-designed three‑tier blank strategy. The study underlines how essential “blank” wells and buffer‑only injections are in single‑cell workflows. conceptually sound and more thorough than many SCP papers.

Main body

Lines 53-56 –  The paragraph nicely introduces saliva and tear fluid, but no tear-fluid proteomics citation is provided—please add at least one landmark study. More broadly, the biofluid context is thin: only saliva references are given later, and even those focus on bulk proteomics. Consider adding a concise overview of previous saliva- and tear-proteome investigations (bulk and, if any, single-cell) to frame the knowledge gap you aim to fill. Finally, spell out—up front—whether the primary goal is analytical-workflow development, biological discovery, or both; at present that distinction is easy to miss.

Results

Figure 1.

Fig 1.b-Add replicate points overlaid on the bars (n =?). –Because injection volumes vary from 50 to 200 nL in the current study (line 624) and the Vanquish Neo specifies >0.25 % precision below 2 µL injections.

-Blank injections (n = 6) are only shown in Fig.1.e but not in dilution‑series data. Please include blank runs for the 50 pg HeLa curve to document carry‑over and false‑positive rate at the lowest load.

- Please add definitions of “method evaluation” feature in Spectronaut”. On an additional note, an independent DIA-NN search will help clarity.

Lines 97-101 (Fig 1 c,f) MS1 and MS2 level of CV calculation was not clear. Is this a feature within Spectronaut that was not explained?

Fig 1.e -Cleaning‑step optimization is said to improve single‑cell CVs, yet the protocol change is buried in Methods. Please summarize the change in the caption and quantify its effect.

Lines 110-112 - Choice of peptides is not justified—are they representative of sample prep digestion efficiency, separation column performance, or etc? Are Supplementary Fig 2B and Supplementary  Fig 3A peptides the same ones?

Lines 116-117  -The text gives the SD for VATVSLPR but omits LSSPATLNSR. Readers cannot tell whether both markers behaved similarly.

Line 119  -median peak width given yet lacks the important fact of points‑per‑peak  to see if the method is quantitative.

Line 123  - FAIMS rationale is speculative- only one FAIMS CV was experimented?

Line 144 -Criteria for “optimization” unclear, how and why for the steps.  Blank 2 and Blank 3 results are absent. Only Blank 1 metrics are reported. Readers cannot judge baseline contamination or LC carry‑over.

Line 224 - The description of saliva and tear cell data processing is confusing.

This raises questions about how thresholds were applied and why discrepancies occurred.

Line 230  Similar to line 224, this explanation lacks clarity. The rationale behind switching thresholds and reprocessing steps needs more justification.

Fig 2. It seems the authors are claiming that excluding certain files led to more protein identifications in the remaining files (Figure 2a/b). If so, why would excluding poor-quality data improve identification in other files? This point needs clarification.

Supplemental Fig. 3 is labeled as evaluating “trypsin efficiency,” but only shows retention times for two trypsin autolysis peptides. Retention time alone doesn’t reflect enzymatic efficiency.

Fig. 4, the proteins should be labeled by gene name for interpretability.

In Fig. 4C/D, there are five Leiden clusters shown, but only four in Figure 4E/F—why the discrepancy?

Supplemental Fig. 6a and 6b appear nearly identical; it’s unclear how they support the point being made.

Fig 5. It is unclear of the usage of a Venn diagram in Fig 5.a. What is the explanation of the drugs detected in both saliva and tear? Any biological pathways? Warrants more explanation.

Line 379-   UMAP based on GO terms is novel, but unsure how informative or valid this approach is.

-the comparison of clusters by GO frequency counts appears novel as opposed to the more well known Leiden approach. could the authors discuss more on what information each approach provides? for eg-, if GO results were compared between Leidens, are they similar to results from the first approach?

Methods

Lines 629- the use of “tryptic protein digestion” in column conditioning is unclear.

Line 647-  Please specify what criteria qualified volunteers as “healthy.” Were smokers, regular alcohol consumers, or individuals taking medication excluded? Also clarify the sampling regimen: at what time of day were saliva and tear fluids collected, and under what fasting conditions (e.g., one hour after the last meal, or after an overnight fast before breakfast)?

Line 658-681-  Biofluids collection- No inhibitors added during saliva/tear collection?

Line 743-  The optimized cleaning protocol is a useful technical contribution. In particular, the inclusion of a “Sterilization” step using 0.5% sodium hypochlorite, 3% hydrogen peroxide, and 70% ethanol, in addition to the standard 70% ethanol wash, appears to improve protein group detection versus fiber diameter. What do each of these reagents do? What is the rationale behind using the reagents and steps?

Line 669/675 -  duplicated lines

Line 756 -  AF be FA?

Biological Interpretation

Please provide clearer rationale and supporting evidence for your cell-type assignments. The claim that neurons were detected in tears seems unlikely and warrants further validation. A marker that is shared across adipocytes, endothelial cells, monocytes, and neutrophils is unlikely to be specific or useful for distinguishing cell types.

The comparison of this data with FDA-validated targets is a nice demonstration of the potential utility of the technique. Are there citations of previous successful therapeutic monitoring of these or similar targets in tears or saliva? Do tears and saliva reflect drug effects through similar or orthogonal pathways?

Competing interests

The authors declare that they have no competing interests.