Solubility-Weighted Index: fast and accurate prediction of protein solubility
Open accessPublished version of a preprint
The DOI matches its Crossref record; 82 of the 85 references it cites were found in a registry or an index; 6 of the 17 sources its claims cite were checked for support.
Abstract
Recombinant protein production is a widely used technique in the biotechnology and biomedical industries, yet only a quarter of target proteins are soluble and can therefore be purified. We have discovered that global structural flexibility, which can be modeled by normalized B-factors, accurately predicts the solubility of 12216 recombinant proteins expressed in Escherichia coli. We call this new predictor the 'Solubility-Weighted Index' (SWI). SWI outperforms many existing protein solubility prediction tools.
Overview
CoChat Summary
TL;DR
The paper introduces the Solubility-Weighted Index (SWI), a sequence-composition score derived from optimized normalized B-factors, to predict recombinant protein solubility in *E. coli*. On the PSI: Biology dataset of 12,216 proteins, SWI improved prediction over flexibility-based scoring and outperformed most compared tools, while also supporting a web server, SoDoPE, for region-level solubility prediction.
Key findings
- Protein structural flexibility was the strongest among the tested sequence properties, with solubility prediction AUC = 0.67 on the PSI: Biology dataset.
- Optimizing normalized B-factors yielded SWI, with training AUC = 0.72 ± 0.00 and test AUC = 0.71 ± 0.01, described as a 7.5% improvement over flexibility in solubility prediction.
- On the independent eSOL dataset (N = 3,198), correlation with solubility improved from Spearman’s rho = 0.40 for the initial B-factors to 0.50 for the final weights; removing His-tags reduced it slightly to rho = 0.47.
- SWI outperformed other compared solubility tools except Protein-Sol for *E. coli* solubility prediction, and the authors report that their SWI C program was the fastest algorithm tested.
Methods
- The main dataset was 12,216 PSI: Biology recombinant proteins expressed in *E. coli* (from DNASU), with validation on the independent eSOL *E. coli* dataset (N = 3,198) and a stickiness dataset of 397 proteins.
- The authors evaluated 9 standard and 9,920 miscellaneous sequence properties, then optimized normalized B-factors using 10-fold cross-validation, 10 clusters of roughly 1,200 sequences, 1,000 bootstrap resamplings per fold, and Nelder–Mead optimization targeting AUC.
- They benchmarked SWI against scalable existing predictors and also fit a logistic regression to convert SWI into a solubility probability for the SoDoPE web server.
Limitations
- The paper states that the SWI probability model should be used only when the protein is successfully expressed in *E. coli*.
- The authors note a possible limitation that SWI may overfit to His-tag fusion proteins, based on the slight drop in eSOL correlation after His-tag removal.
- Evident from the text: the study focuses on *E. coli* expression systems and the compared datasets, so the findings may not generalize to other hosts or solubility contexts without additional validation.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures




Verification
Claims the paper makes
5 Valid · 0 Not valid · 4 Not checked · 44 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Amino acid residues with a lower aromaticity or hydrophilic are known to improve protein solubility (Han et al., 2019; Kramer et al., 2012; Niwa et al., 2009; Trevino et al., 2007; Warwicker et al., 2014; Wilkinson and Harrison, 1991).”
§ Solubility-Weighted Index: fast and accurate prediction of protein solubility > 4 Discussion
What the sources say
- Not matchedcites Han et al., 2019
An author–year citation the paper’s reference list doesn’t settle. We don’t search these by name.
- Not checked — no full text of the sourcecites Kramer et al., 2012 matched to “Toward a molecular understanding of protein solubility: increased negative surface charge correlates with increased solubility” (2012) source
- Partly supportedcites Niwa et al., 2009 matched to “Bimodal protein solubility distribution revealed by an aggregation analysis of the entire ensemble of Escherichia coli proteins” (2009) sourceExperimental — accuracy not yet measured
Niwa et al. (2009) support only the aromaticity half: in 'Relationship Between Solubility and Physicochemical Properties' they report that higher contents of aromatic residues (Phe, Tyr, Trp) were slightly biased toward aggregation-prone, so 'Asp/Glu-rich and/or aromatic-poor proteins tend to be soluble.' The hydrophilic/hydrophobicity half is not supported and is in fact contrary to their findings — the same section states that no significant difference was observed in hydrophobic residue content (Val, Leu, Ile) and that hydropathy-plot and hydrophobic-polar alternate analyses 'failed' to detect any bias with solubility. Their soluble-associated signal is charge (Asp/Glu) and low pI, not hydrophilicity. (LLM, conf=0.82)
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
Every source the paper cites for this claim was checked and supports it.
In the paper
“To validate the cross-validation results, we used a dataset independent of the PSI: Biology known as eSOL (Niwa et al., 2009) (Supplementary Table S1B).”
§ Solubility-Weighted Index: fast and accurate prediction of protein solubility > 3 Results > 3.2 The SWI is an improved predictor of solubility
What the sources say
- Supportedcites Niwa et al., 2009 matched to “Bimodal protein solubility distribution revealed by an aggregation analysis of the entire ensemble of Escherichia coli proteins” (2009) source
The citation is an identification one, pointing to the source of a protein-solubility dataset. Our copy is exactly that work: it reports chaperone-free solubility measurements for the entire E. coli ORF ensemble (3,173 quantified proteins, Agg <30% / Sol >70%; 'Bimodal Distribution of Protein Solubility', and Fig. 2 legend), with per-protein values tabulated (Table S1, 'Relationship Between Solubility and Physicochemical Properties'), and frames the output as a resource/database for aggregation-propensity prediction (Introduction and Discussion). Our copy never prints the label 'eSOL' and never mentions PSI:Biology, but the independence-from-PSI:Biology statement is the citing authors' own description of their validation design rather than a claim about this paper's content. (LLM, conf=0.80)
plainn = 3,198- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“suggesting that solubility is modulated by cotranslational folding (Davis et al., 1999; Natan et al., 2018).”
§ Solubility-Weighted Index: fast and accurate prediction of protein solubility > 3 Results > 3.2 The SWI is an improved predictor of solubility
What the sources say
- Not checked — no full text of the sourcecites Davis et al., 1999 matched to “New fusion protein systems designed to give soluble expression inEscherichia coli” (1999) doi:10.1002/(sici)1097-0290(19991120)65:4
- Supportedcites Natan et al., 2018 matched to “Cotranslational protein assembly imposes evolutionary constraints on homomeric proteins” (2018) source
The Introduction (page unavailable) says cotranslational folding can protect proteins from aggregation. The Results section on engineered constructs (page unavailable) reports that changing the position of an oligomerization domain changed protein solubility in a translation-dependent assembly setting. Together, these passages support the citing sentence’s hedged suggestion that cotranslational folding modulates solubility. (LLM, conf=0.87)
hedgedstates a possible causen = 348- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, a low correlation was obtained upon convergence (Spearman’s rho = 0.18, P = 7.20×10−4; Supplementary Fig. S6B).”
§ Solubility-Weighted Index: fast and accurate prediction of protein solubility > 3 Results > 3.2 The SWI is an improved predictor of solubility
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our solubility scoring using the final weights showed a significant improvement in correlation with E.coli protein solubility over the initial weights (Smith et al.’s normalized B-factors) [Spearman’s rho of 0.50 (P = 2.51×10−205) versus 0.40 (P = 4.57×10−120)].”
§ Solubility-Weighted Index: fast and accurate prediction of protein solubility > 3 Results > 3.2 The SWI is an improved predictor of solubility
plainstates an associationn = 3,198- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
38 more claims
Create a free account to see all 53 claims and to verify the 2 references still open.
- Probably not a substantial claim. 6
- It may state two separate claims as one. 4
- The quote only partly states this claim. 1
Citations
Cites
Scikit-learn: Machine Learning in Python
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Müller, A., Nothman, J., Louppe, G., et al. (2012) · arXiv · doi:10.48550/arxiv.1201.0490 · Cited by 71,016
Matplotlib: a 2D graphics environment
Hunter, J. D. (2007) · Computing in Science & Engineering · doi:10.1109/mcse.2007.55 · Cited by 39,249
A Simplex Method for Function Minimization
Nelder, J. A., & Mead, R. (1965) · The Computer Journal · doi:10.1093/comjnl/7.4.308 · Cited by 28,905
A simple method for displaying the hydropathic character of a protein
Kyte, J., & Doolittle, R. F. (1982) · Journal of Molecular Biology · doi:10.1016/0022-2836(82)90515-0 · Cited by 22,843
Search and clustering orders of magnitude faster than BLAST
Edgar, R. C. (2010) · Bioinformatics · doi:10.1093/bioinformatics/btq461 · Cited by 20,826
A new coronavirus associated with human respiratory disease in China
Wu, F., Zhao, S., Yu, B., Chen, Y.-M., Wang, W., Song, Z.-G., Hu, Y., Tao, Z.-W., Tian, J.-H., Pei, Y.-Y., et al. (2020) · Nature · doi:10.1038/s41586-020-2008-3 · Cited by 11,688
Data Structures for Statistical Computing in Python
McKinney, W. (2010) · Proceedings of the Python in Science Conference · doi:10.25080/majora-92bf1922-00a · Cited by 11,032
The NumPy Array: A Structure for Efficient Numerical Computation
van der Walt, S., Colbert, S. C., & Varoquaux, G. (2011) · Computing in Science & Engineering · doi:10.1109/mcse.2011.37 · Cited by 9,381
Scikit-learn: Machine Learning in Python
(2011) · Journal of Machine Learning Research · doi:10.5555/1953048.2078195 · Cited by 8,539
Statsmodels: Econometric and Statistical Modeling with Python
Seabold, S., & Perktold, J. (2010) · Proceedings of the Python in Science Conference · doi:10.25080/majora-92bf1922-011 · Cited by 6,465
Research works printed but not linked: 1
- Harrison R.G. (2000) Expression of soluble heterologous proteins via fusion with NusA protein. Innovations, 11, 4– 7.
Cited by
Items 101–105 of 105
Sallah, S., & Warwicker, J. (2024) · doi:10.1101/2024.07.17.603952 · Cited by 3
Behavior of proteins alone and upon interactions with nanosilicas and polymers in aqueous media
Гунько, В. М. (2026) · Himia Fizika ta Tehnologia Poverhni · doi:10.15407/hftp17.03.283
Design and Cloning of Gene Encoding SLPI C-Terminal Domain in Escherichia coli TOP10
Ulfa, E. U., & Puspaningsih, N. N. T. (2022) · Indonesian Journal of Pharmaceutical Science and Technology · doi:10.24198/ijpst.v9i3.36918
Pro4S: prediction of protein solubility by fusing sequence, structure, and surface
Qian, J., Yang, L., Wang, R.-X., & Qi, Y.-F. (2025) · bioRxiv · doi:10.1002/pro.70749 · Cited by 1
Yang, Y., Ma, H., 王东培, Jiang, Z., Wang, X., Zhao, Y., Li, R., & Xu, C. (2026) · Food Research International · doi:10.1016/j.foodres.2026.119535
Metadata
- Authors
- Bhandari, B. K., Gardner, P. P., & Lim, C. S.
- Affiliations
- University of Otago (NZ)
- University of Canterbury (NZ)
- Corresponding author
- Lim, Chun Shen
- Venue
- Bioinformatics
- Publisher
- Oxford University Press (OUP)
- Published
- 2020-06-19
- Volume
- 36
- Issue
- 18
- Pages
- 4691-4698
- ISSN
- 1367-4803 (ISSN-L)
- 1367-4811
- PMC
- PMC7750957
- OpenAlex
- W3036344389
- Semantic Scholar
- 8b415f71f4f52187a70e94e8e064d15aa9975f3d
- Cited by
- 97
- References
- 85
- Citations by year
- 2026: 17 · 2025: 19 · 2024: 23 · 2023: 11 · 2022: 9 · 2021: 13 · 2020: 5
- FWCI
- 3.16
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Bioinformatics (publishedVersion, cc-by)
- bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Funders
- Ministry of Business, Innovation and Employment, New Zealand [MBIE (UOOX1709)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein purification and stability
- Keywords
- protein solubility prediction, structural flexibility, recombinant protein expression, Escherichia coli expression, protein expression optimization, protein purification
- MeSH terms
- Computers, Escherichia coli, Proteins, Software, Solubility
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash and gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.