Search

A deep learning functional for the 1D Reference Interaction Site Model (1D RISM) enables accurate predictions of solvation entropy, enthalpy and free energy.

pyRISM combines physics-based calculations and deep learning to rapidly predict solvation free energy in different solvents and temperatures without reparameterization.

Attenuated total reflectance (ATR)-Fourier transform infrared (FTIR) spectroscopy alongside machine learning (ML) techniques is an emerging approach for the early detection of brain cancer in clinical practice. A crucial step in the acquisition of an IR spectrum is the transformation of the time domain signal from the biological sample to a frequency domain spectrum via a discrete Fourier transform. Further pre-processing of the spectrum is typically applied to reduce non-biological sample variance, and thus to improve subsequent analysis. However, the Fourier transformation is often assumed to be essential even though modelling of time domain data is common in other fields. We apply an inverse Fourier transform to frequency domain data to map these to the time domain. We use the transformed data to develop deep learning models utilising Recurrent Neural Networks (RNNs) to differentiate between brain cancer and control in a cohort of 1438 patients. The best performing model achieves a mean (cross-validated score) area under the receiver operating characteristic (ROC) curve (AUC) of 0.97 with sensitivity of 0.91 and specificity of 0.91. This is better than the optimal model trained on frequency domain data which achieves an AUC of 0.93 with sensitivity of 0.85 and specificity of 0.85. A dataset comprising 385 patient samples which were prospectively collected in the clinic is used to test a model defined with the best performing configuration and fit to the time domain. Its classification accuracy is found to be comparable to the gold-standard for this dataset demonstrating that RNNs can accurately classify disease states using spectroscopic data represented in the time domain.

Pancreatic cancer claims over 460,000 victims per year. The carbohydrate antigen (CA) 19-9 test is the blood test used for pancreatic cancer’s detection; however, its levels can be raised in symptomatic patients with other non-malignant diseases, or with other tumors in the surrounding area. Attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy has demonstrated exceptional potential in cancer diagnostics, and its clinical implementation could represent a significant step towards early detection. This proof-of-concept study, investigating the use of ATR-FTIR spectroscopy on dried blood serum, focused on the discrimination of both cancer versus healthy control samples, and cancer versus symptomatic non-malignant control samples, as a novel liquid biopsy approach for pancreatic cancer diagnosis. Machine learning algorithms were applied, achieving results of up to 92% sensitivity and 88% specificity when discriminating between cancers (n = 100) and healthy controls (n = 100). An area under the curve (AUC) of 0.95 was obtained through receiver operating characteristic (ROC) analysis. Balanced sensitivity and specificity over 75%, with an AUC of 0.83, were achieved with cancers (n = 35) versus symptomatic controls (n = 35). Herein, we present these results as demonstration that our liquid biopsy approach could become a simple, minimally invasive, and reliable diagnostic test for pancreatic cancer detection.

A rapid, low-cost, sensitive, multi-cancer early detection (MCED) test would be transformational in the diagnostics field. Earlier cancer detection can increase survival rates and quality of life of patients. An effective test must accurately identify the small proportion of patients with typically non-specific symptoms who have cancer. Such symptoms do not easily segregate by organ system, necessitating a multi-cancer approach. In this large-scale study (n = 2094 patients) we applied the Dxcover® Cancer Liquid Biopsy to differentiate cancer against non-cancer patients, as well as organ specific tests to identify cancers of the brain, breast, colorectal, kidney, lung, ovary, pancreas, and prostate. The test uses Fourier transform infrared (FTIR) spectroscopy to analyze all macromolecules in a minute volume of patient serum, and machine learning algorithms to build a classifier of the resultant spectral profiles to detect cancer. This approach can be fine-tuned to maximize either sensitivity or specificity depending on the requirements from different healthcare systems and cancer diagnostic pathways. The cancer v asymptomatic non-cancer classification detected 64% of stage I cancers when specificity was 99% (overall sensitivity 56%). When tuned for higher sensitivity, this model identified 99% of stage I cancers (while specificity was 58%). When examining cancer against all non-cancer (including symptomatic patients), the sensitivity-tuned model enabled 90% sensitivity with 61% specificity, with detection rates of 93% for stage I, 84% for stage II, 92% for stage III and 95% for stage IV. For organ specific cancer classifiers, area under the receiver operating characteristic (ROC) curve values were calculated for all cancers: brain (0.90), breast (0.75), colorectal (0.91), kidney (0.91), lung (0.91), ovarian (0.86), pancreatic (0.85) and prostate (0.86). Cancer treatment is more effective when given earlier and this low-cost strategy can facilitate the requisite earlier diagnosis.

D.S.P. and D.J.F. thank the EPSRC for funding via Prosperity Partnership EP/S035990/1. D.S.P. and D.J.F. thank the ARCHIE-WeSt High-Performance Computing Centre (www.archie-west.ac.uk) for computational resources. The UCL authors thank Prof. Keith Refson for guidance with the phonon calculations, which used the ARCHER U.K. National Supercomputing Service (http://www.archer.ac.uk) as part of the U.K. HEC Materials Chemistry Consortium, which is funded by the EPSRC (EP/L000202, EP/R029431). R.G. was funded by MagnaPharm, a project funded by the European Union’s Horizon 2020 Research and Innovation programme under grant agreement number 736899. We demonstrate that physics-based calculations of intrinsic aqueous solubility can rival cheminformatics-based machine learning predictions. A proof-of-concept was developed for a physics-based approach via a sublimation thermodynamic cycle, building upon previous work that relied upon several thermodynamic approximations, notably the 2RT approximation, and limited conformational sampling. Here, we apply improvements to our sublimation free-energy model with the use of crystal phonon mode calculations to capture the contributions of the vibrational modes of the crystal. Including these improvements with lattice energies computed using the model-potential-based Ψmol method leads to accurate estimates of sublimation free energy. Combining these with hydration free energies obtained from either molecular dynamics free-energy perturbation simulations or density functional theory calculations, solubilities comparable to both experiment and informatics predictions are obtained. The application to coronene, succinic acid, and the pharmaceutical desloratadine shows how the methods must be adapted for the adoption of different conformations in different phases. The approach has the flexibility to extend to applications that cannot be covered by informatics methods. Publisher PDF

Background: Early detection of cancer is vital to improve patient survival, since earlier diagnosis and treatment maximizes the opportunity to combat or control disease progression. Specifically for colorectal cancer (CRC), the average 5-survival rate after diagnosis decreases from 91% in early-stage CRC, to as low as 15% for stage IV CRC. Furthermore, the rapid detection and subsequent removal of pre-cancerous adenomas - e.g., advanced adenomas (AA) - can improve survival rates of affected patients. Current stool-based tests, such as FIT and FIT-DNA, have poor AA sensitivities of 24% and 42%, and specificities of 95% and 87%, respectively. Recent research into liquid biopsies have shown some promise, however tests based on tumor-derived biomarkers alone have limited sensitivity, especially in early-stage disease. Methods: We initially applied the Dxcover® Cancer Liquid Biopsy for use as a multi-cancer early detection (MCED) test. The test uses Fourier transform infrared (FTIR) spectroscopy and machine learning algorithms to build a classifier of the resultant spectral profiles to detect cancer, and can be fine-tuned to maximize either sensitivity or specificity depending on the requirements of specific international healthcare systems. Additionally, we have analyzed a retrospective cohort of serum samples comprising 100 CRC, 99 advanced adenomas removed by surgical resection and 97 colonoscopy screening controls. Results: The CRC classifier from the discovery MCED dataset reported an area under the receiver operating characteristic curve (AUROC) value of 0.91, with 74% sensitivity and 91% specificity when differentiating CRC and non-cancer, which surpasses the targets set by the Centers for Medicare & Medicaid Services (CMS) for coverage of CRC tests. When tuned for higher sensitivity, the model produced 97% sensitivity (49% specificity), and when tailored for greater specificity (97%) the sensitivity was 47%. In this study, we have progressed these findings to examine the ability of the technology to differentiate patients with CRC, AA and colonoscopy controls. Conclusions: Cancer treatment is more effective when given earlier and this low-cost strategy can facilitate the requisite earlier diagnosis. A rapid blood test that sensitive to AA and early-stage CRC could improve patient prognosis and ultimately reduce mortality. Citation Format: James M. Cameron, Georgios Antoniou, Paul M. Brennan, Justin J. Conn, Sharon King, Rose G. McHardy, Susan Moug, Jennifer Nobes, David S. Palmer, Alexandra Sala, Benjamin R. Smith, Judith Strachan, Craig Mowat, Matthew J. Baker. Early colorectal cancer detection with a spectroscopic liquid biopsy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6506.

Identification of correct protein–ligand binding poses is important in structure-based drug design and crucial for the evaluation of protein–ligand binding affinity. Protein–ligand coordinates are commonly obtained from crystallography experiments that provide a static model of an ensemble of conformations. Binding pose metadynamics (BPMD) is an enhanced sampling method that allows for an efficient assessment of ligand stability in solution. Ligand poses that are unstable under the bias of the metadynamics simulation are expected to be infrequently occupied in the energy landscape, thus making minimal contributions to the binding affinity. Here, the robustness of the method is studied using crystal structures with ligands known to be incorrectly modeled, as well as 63 structurally diverse crystal structures with ligand fit to electron density from the Twilight database. Results show that BPMD can successfully differentiate compounds whose binding pose is not supported by the electron density from those with well-defined electron density.

Background Diagnostic delays impact the quality of life and survival of patients with brain tumors. Earlier and expeditious diagnoses in these patients are crucial to reduce the morbidities and mortalities associated with brain tumors. A simple, rapid blood test that can be administered easily in a primary care setting to efficiently identify symptomatic patients who are most likely to have a brain tumor would enable quicker referral to brain imaging for those who need it most. Methods Blood serum samples from 603 patients were prospectively collected and analyzed. Patients either had non-specific symptoms that could be indicative of a brain tumor on presentation to the Emergency Department, or a new brain tumor diagnosis and referral to the neurosurgical unit, NHS Lothian, Scotland. Patient blood serum samples were analyzed using the Dxcover® Brain Cancer liquid biopsy. This technology utilizes infrared spectroscopy combined with a diagnostic algorithm to predict the presence of intracranial disease. Results Our liquid biopsy approach reported an area under the receiver operating characteristic curve of 0.8. The sensitivity-tuned model achieves a 96% sensitivity with 45% specificity (NPV 99.3%) and identified 100% of glioblastoma multiforme patients. When tuned for a higher specificity, the model yields a sensitivity of 47% with 90% specificity (PPV 28.4%). Conclusions This simple, non-invasive blood test facilitates the triage and radiographic diagnosis of brain tumor patients while providing reassurance to healthy patients. Minimizing time to diagnosis would facilitate the identification of brain tumor patients at an earlier stage, enabling more effective, less morbid surgical and adjuvant care.

Simple Summary Due to the non-specific symptoms of brain cancer (e.g., headaches or memory changes), gliomas will often remain undetected until they are larger or at a higher grade, reducing the patient’s likelihood of a good clinical outcome. Earlier detection and diagnosis of brain tumours is vital to improve patient outcomes, leading to safer surgeries and earlier treatments. A liquid biopsy for brain tumour would prove revolutionary however in order to detect disease earlier the liquid biopsy needs to be able to detect smaller tumours; and current liquid biopsies perform worse when detecting smaller or earlier stage tumours. Here, for the first time, we confirm the applicability of a validated spectroscopic liquid biopsy approach to detect both small and low-grade gliomas proving that the spectroscopic liquid biopsy approach is insensitive to tumour volume unlike other liquid biopsies. Abstract Background: To support the early detection and diagnosis of brain tumours we have developed a rapid, cost-effective and easy to use spectroscopic liquid biopsy based on the absorbance of infrared radiation. We have previously reported highly sensitive results of our approach which can discriminate patients with a recent brain tumour diagnosis and asymptomatic controls. Other liquid biopsy approaches (e.g., based on tumour genetic material) report a lower classification accuracy for early-stage tumours. In this manuscript we present an investigation into the link between brain tumour volume and liquid biopsy test performance. Methods: In a cohort of 177 patients (90 patients with high-grade glioma (glioblastoma (GBM) or anaplastic astrocytoma), or low-grade glioma (astrocytoma, oligoastrocytoma and oligodendroglioma)) tumour volumes were calculated from magnetic resonance imaging (MRI) investigations and patients were split into two groups depending on MRI parameters (T1 with contrast enhancement or T2/FLAIR (fluid-attenuated inversion recovery)). Using attenuated total reflection (ATR)-Fourier transform infrared (FTIR) spectroscopy coupled with supervised learning methods and machine learning algorithms, 90 tumour patients were stratified against 87 control patients who displayed no symptomatic indications of cancer, and were classified as either glioma or non-glioma. Results: Sensitivities, specificities and balanced accuracies were all greater than 88%, the area under the curve (AUC) was 0.98, and cancer patients with tumour volumes as small as 0.2 cm3 were correctly identified. Conclusions: Our spectroscopic liquid biopsy approach can identify gliomas that are both small and low-grade showing great promise for deployment of this technique for early detection and diagnosis.

To support the early detection and diagnosis of brain tumors we have developed a rapid, cost-effective and easy to use spectroscopic liquid biopsy based on the absorbance of infrared radiation. We have previously reported highly sensitive results of our approach which can discriminate patients with a recent brain tumor diagnosis and asymptomatic controls. Other liquid biopsy approaches (e.g., based on tumor genetic material) report a lower classification accuracy for early-stage tumors. In this manuscript we present an investigation into the link between brain tumor volume and liquid biopsy test performance. In a cohort of 177 patients (90 patients with high-grade glioma (glioblastoma (GBM) or anaplastic astrocytoma), or low-grade glioma (astrocytoma, oligo-astrocytoma and oligodendroglioma)) tumor volumes were calculated from magnetic resonance imaging (MRI) investigations and patients were split into two groups depending on MRI parameters (T1 with contrast enhancement or T2/FLAIR (fluid-attenuated inversion recovery)). Using attenuated total reflection (ATR)-Fourier transform infrared (FTIR) spectroscopy coupled with supervised learning methods and machine learning algorithms, 90 tumor patients were stratified against 87 control patients who displayed no symptomatic indications of cancer, and were classified as either glioma or non-glioma. Sensitivities, specificities and balanced accuracies were all greater than 88%, the area under the curve (AUC) was 0.98, and cancer patients with tumor volumes as small as 0.2 cm3 were correctly identified. Our spectroscopic liquid biopsy approach can identify gliomas that are both small and low-grade showing great promise for deployment of this technique for early detection and diagnosis. Citation Format: Alexandra Sala, Ashton G. Theakstone, Paul M. Brennan, Michael D. Jenkinson, Samantha J. Mills, Khaja Syed, Christopher Rinaldi, Yun Xu, Royston Goodacre, Holly J. Butler, David S. Palmer, Benjamin R. Smith, Matthew J. Baker. Rapid spectroscopic liquid biopsy for the early detection of brain tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5923.

Diagnostic delays impact the quality of life and survival of patients with brain tumors. Earlier and expeditious diagnoses in these patients are crucial to reducing the morbidities and mortalities associated with brain tumors. A simple, rapid blood test that can be administered easily in a primary care setting to efficiently identify symptomatic patients who are most likely to have a brain tumor would enable quicker referral to brain imaging for those who need it most. Blood serum samples from 603 patients were prospectively collected and analyzed. Patients either had non-specific symptoms that could be indicative of a brain tumor on presentation to the Emergency Department, or a new brain tumor diagnosis and referral to the neurosurgical unit, NHS Lothian, Scotland. Patient blood serum samples were analyzed using the Dxcover® Brain Cancer liquid biopsy. This technology utilizes infrared spectroscopy combined with a diagnostic algorithm to predict the presence of intracranial disease. Our liquid biopsy approach reported an area under the receiver operating characteristic curve of 0.8. The sensitivity-tuned model achieves a 96% sensitivity with 45% specificity (NPV 99.3%) and identified 100% of glioblastoma multiforme patients. When tuned for a higher specificity, the model yields sensitivity of 47% with 90% specificity (PPV 28.4%). This simple, non-invasive blood test facilitates the triage and radiographic diagnosis of brain tumor patients, while providing reassurance to healthy patients. Minimizing time to diagnosis would facilitate identification of brain tumor patients at an earlier stage, enabling more effective, less morbid surgical and adjuvant care. Citation Format: James M. Cameron, Paul M. Brennan, Georgios Antoniou, Holly J. Butler, Loren Christie, Justin J.A. Conn, Tom Curran, Ewan Gray, Mark G. Hegarty, Michael Jenkinson, Daniel Orringer, David S. Palmer, Alexandra Sala, Benjamin R. Smith, Matthew J. Baker. Clinical validation of a spectroscopic liquid biopsy for early detection of brain cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5922.

Pancreatic cancer is the 7th most deadly cancer worldwide with over 460,000 victims per year. In the current diagnostic pathway, carbohydrate antigen (CA) 19-9 serum test is the first method used for detection of pancreatic cancer; although, with poor positive predictive values reported, it cannot provide certain information about the presence of a pancreatic tumor or other surrounding tumors in symptomatic patients. Attenuated total reflection - Fourier transform infrared spectroscopy (ATR-FTIR) has demonstrated exceptional potential in human blood serum analysis for cancer diagnostics and its implementation in the clinical environment could represent a significant step forward in the early detection of pancreatic cancer. This proof-of-concept study was focused on the discrimination between both cancer versus healthy control samples, and cancer versus symptomatic control samples from patients with comorbidities and/or confounding diseases. The study aimed to investigate the use of the Dxcover® cancer platform as a novel approach for pancreatic cancer detection. Various machine learning algorithms were applied to discriminate between cancer (n=100) and healthy control samples (n=100), achieving results amounting to a sensitivity of 91.0 ± 5.6%, specificity of 87.6 ± 5.8%, and accuracy of 89.3 ± 3.8% with partial least squares-discriminant analysis (PLS-DA). Moreover, an area under the curve (AUC) equal to 0.9536 was obtained through receiving operating characteristic (ROC) analysis. The same algorithms were also applied to discriminate between cancer (n=35) and symptomatic control samples (n=35) achieving a balanced sensitivity and specificity over 75% with an AUC of 0.8436. Both discriminations underwent bootstrapping validation and were proven statistically significant. Herein, we present these results and demonstrate that ATR-FTIR spectroscopic analysis of serum is a cost-effective, minimally invasive, highly sensitive and specific test for detection of pancreatic cancer. Citation Format: Alexandra Sala, James M. Cameron, Cerys A. Jenkins, Hugh Barr, Loren Christie, Justin J. Conn, Thomas (Jeff) R. Evans, Dean A. Harris, David S. Palmer, Christopher Rinaldi, Ashton G. Theakstone, Matthew J. Baker. Machine learning based detection of pancreatic tumors using the Dxcover® cancer platform [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5921.

The development of a robust multi-cancer early detection (MCED) test would be transformational within the cancer diagnostics field, as earlier detection is vital to inhibit disease progression, improve patient prognosis and reduce mortality rates. In this large-scale spectroscopic study, we present the Dxcover® Cancer Liquid Biopsy, the world’s first infrared spectroscopy-based blood test for the early detection of multiple cancers. The full cohort (n=2094) was comprised of patients with a confirmed cancer diagnosis (n=1543), symptomatic non-cancer patients (n=460) and healthy volunteers (n=91). For the cancer versus non-cancer classification, receiver operating characteristic analysis produced a mean curve with an area under the curve (AUC) value of 0.86. When adjusted for greater sensitivity, our model achieved 90% sensitivity with 61% specificity, and yielded a sensitivity of 56% with 91% specificity when tailored for higher specificity. We also tuned for maximized sensitivity or specificity, whilst the other parameter was fixed above a minimum value of 45% for the cross-validation result, based upon performance levels of currently available clinically benchmarked tests. This resulted in a 94% sensitivity where specificity was 47%, and a 94% specificity with 48% sensitivity. The overall detection rates were 93% for stage I, 84% for stage II, 92% for stage III and 95% for stage IV for the sensitivity-tuned model. The specificity-tuned model detected 90% of non-cancer patients accurately, resulting in few false negatives (54/551). For the binary classifiers of individual cancer types against symptomatic control patients, promising AUC values were reported for all cancers: brain (0.90), breast (0.74), colorectal (0.91), kidney (0.91), lung (0.90), ovarian (0.85), pancreatic (0.81) and prostate (0.85). Importantly, the detection rates did not vary significantly according to cancer stage, suggesting the technology is very effective at detecting early-stage tumors, which is a necessity for MCED tests. Citation Format: James M. Cameron, Alexandra Sala, Georgios Antoniou, Paul M. Brennan, Justin J. A. Conn, Siobhan Connal, David S. Palmer, Benjamin R. Smith, Matthew J. Baker. Multi-cancer early detection with a spectroscopic liquid biopsy platform [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5920.

Early diagnosis of brain tumours is challenging and a major unmet need. Patients with brain tumours most often present with non-specific symptoms more commonly associated with less serious diagnoses, making it difficult to determine which patients to prioritize for brain imaging. Delays in diagnosis affect timely access to treatment, with potential impacts on quality of life and survival. A test to help identify which patients with non-specific symptoms are most likely to have a brain tumour at an earlier stage would dramatically impact on patients by prioritizing demand on diagnostic imaging facilities. This clinical feasibility study of brain tumour early diagnosis was aimed at determining the accuracy of our novel spectroscopic liquid biopsy test for the triage of patients with non-specific symptoms that might be indicative of a brain tumour, for brain imaging. Patients with a suspected brain tumour based on assessment of their symptoms in primary care can be referred for open access CT scanning. Blood samples were prospectively obtained from 385 of such patients, or patients with a new brain tumour diagnosis. Samples were analysed using our spectroscopic liquid biopsy test to predict presence of disease, blinded to the brain imaging findings. The results were compared to the patient’s index brain imaging delivered as per standard care. Our test predicted the presence of glioblastoma, the most common and aggressive brain tumour, with 91% sensitivity, and all brain tumours with 81% sensitivity, and 80% specificity. Negative predictive value was 95% and positive predictive value 45%. The reported levels of diagnostic accuracy presented here have the potential to improve current symptom-based referral guidelines, and streamline assessment and diagnosis of symptomatic patients with a suspected brain tumour., Brennan et al. show a rapid, low-cost blood test can identify which patients with suspected brain tumour to prioritize for diagnostic imaging, reducing time to diagnosis. The test is more than 90% sensitive for the most aggressive tumours. Earlier diagnosis can provide health and economic benefits., Graphical Abstract Graphical Abstract

Objectives An early economic evaluation to inform the translation into clinical practice of a spectroscopic liquid biopsy for the detection of brain cancer. Two specific aims are (1) to update an existing economic model with results from a prospective study of diagnostic accuracy and (2) to explore the potential of brain tumor-type predictions to affect patient outcomes and healthcare costs. Methods A cost-effectiveness analysis from a UK NHS perspective of the use of spectroscopic liquid biopsy in primary and secondary care settings, as well as a cost–consequence analysis of the addition of tumor-type predictions was conducted. Decision tree models were constructed to represent simplified diagnostic pathways. Test diagnostic accuracy parameters were based on a prospective validation study. Four price points (GBP 50-200, EUR 57-228) for the test were considered. Results In both settings, the use of liquid biopsy produced QALY gains. In primary care, at test costs below GBP 100 (EUR 114), testing was cost saving. At GBP 100 (EUR 114) per test, the ICER was GBP 13,279 (EUR 15,145), whereas at GBP 200 (EUR 228), the ICER was GBP 78,300 (EUR 89,301). In secondary care, the ICER ranged from GBP 11,360 (EUR 12,956) to GBP 43,870 (EUR 50,034) across the range of test costs. Conclusions The results demonstrate the potential for the technology to be cost-effective in both primary and secondary care settings. Additional studies of test use in routine primary care practice are needed to resolve the remaining issues of uncertainty—prevalence in this patient population and referral behavior.

Patients living with brain tumours have the highest average years of life lost of any cancer, ultimately reducing average life expectancy by 20 years. Diagnosis depends on brain imaging and most often confirmatory tissue biopsy for histology. The majority of patients experience non-specific symptoms, such as headache, and may be reviewed in primary care on multiple occasions before diagnosis is made. Sixty-two per cent of patients are diagnosed on brain imaging performed when they deteriorate and present to the emergency department. Histological diagnosis from invasive surgical biopsy is necessary prior to definitive treatment, because imaging techniques alone have difficulty in distinguishing between several types of brain cancer. However, surgery itself does not necessarily control tumour growth, and risks morbidity for the patient. Due to their similar features on brain scans, glioblastoma, primary central nervous system lymphoma and brain metastases have been known to cause radiological confusion. Non-invasive tests that support stratification of tumour subtype would enhance early personalisation of treatment selection and reduce the delay and risks associated with surgery for many patients. Techniques involving vibrational spectroscopy, such as attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy, have previously demonstrated analytical capabilities for cancer diagnostics. In this study, infrared spectra from 641 blood serum samples obtained from brain cancer and control patients have been collected. Firstly, we highlight the capability of ATR-FTIR to distinguish between healthy controls and brain cancer at sensitivities and specificities above 90%, before defining subtle differences in protein secondary structures between patient groups through Amide I deconvolution. We successfully differentiate several types of brain lesions (glioblastoma, meningioma, primary central nervous system lymphoma and metastasis) with balanced accuracies &gt, 80%. A reliable blood serum test capable of stratifying brain tumours in secondary care could potentially avoid surgery and speed up the time to definitive therapy, which would be of great value for both neurologists and patients.

In recent years, the diagnosis of brain tumors has been investigated with attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy on dried human serum samples to eliminate spectral interferences of the water component, with promising results. This research evaluates ATR-FTIR on both liquid and air-dried samples to investigate "digital drying" as an alternative approach for the analysis of spectra obtained from liquid samples. Digital drying approaches, consisting of water subtraction and least-squares method, have demonstrated a greater random forest (RF) classification performance than the air-dried spectra approach when discriminating cancer vs control samples, reaching sensitivity values higher than 93.0% and specificity values higher than 83.0%. Moreover, quantum cascade laser infrared (QCL-IR) based spectroscopic imaging is utilized on liquid samples to assess the implications of a deep-penetration light source on disease classification. The RF classification of QCL-IR data has provided sensitivity and specificity amounting to 85.1% and 75.3% respectively.

The potential to predict Solvation Free Energies (SFEs) in any solvent using a machine learning (ML) model based on thermodynamic output, extracted from 3D-RISM simulations in water is investigated. The models on multiple solvents take into account both the solute and solvent description and offer the possibility to predict SFEs of any solute in any solvent with root mean squared errors less than 1 kcal/mol. Validations that involve exclusion of fractions or clusters of the solutes or solvents exemplify the model’s capability to predict SFEs of novel solutes and solvents with diverse chemical profiles. In addition to being predictive, our models can identify the solute and solvent features that influence SFE predictions. Furthermore, using 3D-RISM hydration thermodynamic output to predict SFEs in any organic solvent reduces the need to run 3D-RISM simulations in all these solvents. Altogether, our multi-solvent models for SFE predictions that take advantage of the solvation effects are expected to have an impact in the property prediction space.

Recently, Gussregen et al. used solute–solvent distribution functions calculated by the three-dimensional Reference Interaction Site Model (3DRISM) in a 3D quantitative structure–activity relationship (QSAR) approach to model activity data for a set of serine protease inhibitors; this approach was referred to as Comparative Analysis of 3D RISM Maps (CARMa). [J. Chem. Inf. Model. 2017, 57, 1652–1666] Here we extend this idea by introducing probe atoms into the 3DRISM solvent model in order to directly capture other molecular interactions in addition to those related to hydration/dehydration. Benchmark results for six different protein–ligand systems show that CARMa models trained on probe atom descriptors give consistently more accurate predictions than Comparative Molecular Field Analysis (CoMFA) and other common QSAR approaches.

Bovine and camel chymosins are aspartic proteases that are used in dairy food manufacturing. Both enzymes catalyze proteolysis of a milk protein, κ-casein, which helps to initiate milk coagulation. Surprisingly, camel chymosin shows a 70% higher clotting activity than bovine chymosin for bovine milk, while exhibiting only 20% of the unspecific proteolytic activity. By contrast, bovine chymosin is a poor coagulant for camel milk. Although both enzymes are marketed commercially, the disparity in their catalytic activity is not yet well understood at a molecular level, due in part to a lack of atomistic resolution data about the chymosin—κ-casein complexes. Here, we report computational alanine scanning calculations of all four chymosin—κ-casein complexes, allowing us to elucidate the influence that individual residues have on binding thermodynamics. Of the 12 sequence differences in the binding sites of bovine and camel chymosin, eight are shown to be particularly important for understanding differences in the binding thermodynamics (Asp112Glu, Lys221Val, Gln242Arg, Gln278Lys. Glu290Asp, His292Asn, Gln294Glu, and Lys295Leu. Residue in bovine chymosin written first). The relative binding free energies of single-point mutants of chymosin are calculated using the molecular mechanics three dimensional reference interaction site model (MM-3DRISM). Visualization of the solvent density functions calculated by 3DRISM reveals the difference in solvation of the binding sites of chymosin mutants.

Over a third of brain tumour patients visit their general practitioner more than five times prior to diagnosis in the UK, leading to 62% of patients being diagnosed as emergency presentations. Unfortunately, symptoms are non-specific to brain tumours, and the majority of these patients complain of headaches on multiple occasions before being referred to a neurologist. As there are currently no methods in place for the early detection of brain cancer, the affected patients' average life expectancy is reduced by 20 years. These statistics indicate that the current pathway is ineffective, and there is a vast need for a rapid diagnostic test. Attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy is sensitive to the hallmarks of cancer, as it analyses the full range of macromolecular classes. The combination of serum spectroscopy and advanced data analysis has previously been shown to rapidly and objectively distinguish brain tumour severity. Recently, a novel high-throughput ATR accessory has been developed, which could be cost-effective to the National Health Service in the UK, and valuable for clinical translation. In this study, 765 blood serum samples have been collected from healthy controls and patients diagnosed with various types of brain cancer, contributing to one of the largest spectroscopic studies to date. Three robust machine learning techniques - random forest, partial least squares-discriminant analysis and support vector machine - have all provided promising results. The novel high-throughput technology has been validated by separating brain cancer and non-cancer with balanced accuracies of 90% which is comparable to the traditional fixed diamond crystal methodology. Furthermore, the differentiation of brain tumour type could be useful for neurologists, as some are difficult to distinguish through medical imaging alone. For example, the highly aggressive glioblastoma multiforme and primary cerebral lymphoma can appear similar on magnetic resonance imaging (MRI) scans, thus are often misdiagnosed. Here, we report the ability of infrared spectroscopy to distinguish between glioblastoma and lymphoma patients, at a sensitivity and specificity of 90.1% and 86.3%, respectively. A reliable serum diagnostic test could avoid the need for surgery and speed up time to definitive chemotherapy and radiotherapy.

Non-specific symptoms, as well as the lack of a cost-effective test to triage patients in primary care, has resulted in increased time-to-diagnosis and a poor prognosis for brain cancer patients. A rapid, cost-effective, triage test could significantly improve this patient pathway. A blood test using attenuated total reflection (ATR)-Fourier transform infrared (FTIR) spectroscopy for the detection of brain cancer, alongside machine learning technology, is advancing towards clinical translation. However, whilst the methodology is simple and does not require extensive sample preparation, the throughput of such an approach is limited. Here we describe the development of instrumentation for the analysis of serum that is able to differentiate cancer and control patients at a sensitivity and specificity of 93.2% and 92.8%. Furthermore, preliminary data from the first prospective clinical validation study of its kind are presented, demonstrating how this innovative technology can triage patients and allow rapid access to imaging., Diagnosing brain cancer is frequently difficult and requires specialist equipment. Here, the authors develop their previous attenuated total reflectance-Fourier transform infrared spectroscopy method and incoporate the use of disposable silicon wafers for diagnosing brain cancer using serum samples.

Simple Summary Gliomas represent the vast majority of primary brain tumours and are of significant medical importance due to the poor clinical course of affected patients. The isocitrate dehydrogenase 1 (IDH1) mutation is associated with improved prognosis, compared to patients with IDH1-wildtype lesions of the same stage. In this proof-of-concept study, Fourier transform infrared spectroscopy was used to determine the IDH1 molecular status in fixed glioma sections. Classification algorithms successfully distinguished the two IDH1 classes with high accuracies (>80%). Knowledge of the IDH1 status would be beneficial, as maximum resection may be preferred in patients with IDH1-mutant gliomas, whilst a more limited resection can be best for IDH1-wildtype gliomas. Furthermore, we examined blood serum in an attempt to identify the biomolecular alterations caused by the IDH1 mutation. Non-invasive approaches that can detect the molecular status may guide some patients to an alternative treatment prior to surgery. Abstract Mutations in the isocitrate dehydrogenase 1 (IDH1) gene are found in a high proportion of diffuse gliomas. The presence of the IDH1 mutation is a valuable diagnostic, prognostic and predictive biomarker for the management of patients with glial tumours. Techniques involving vibrational spectroscopy, e.g., Fourier transform infrared (FTIR) spectroscopy, have previously demonstrated analytical capabilities for cancer detection, and have the potential to contribute to diagnostics. The implementation of FTIR microspectroscopy during surgical biopsy could present a fast, label-free method for molecular genetic classification. For example, the rapid determination of IDH1 status in a patient with a glioma diagnosis could inform intra-operative decision-making between alternative surgical strategies. In this study, we utilized synchrotron-based FTIR microanalysis to probe tissue microarray sections from 79 glioma patients, and distinguished the positive class (IDH1-mutated) from the IDH1-wildtype glioma, with a sensitivity and specificity of 82.4% and 83.4%, respectively. We also examined the ability of attenuated total reflection (ATR)-FTIR spectroscopy in detecting the biomolecular events and global epigenetic and metabolic changes associated with mutations in the IDH1 enzyme, in blood serum samples collected from an additional 72 brain tumour patients. Centrifugal filtration enhanced the diagnostic ability of the classification models, with balanced accuracies up to ~69%. Identification of the molecular status from blood serum prior to biopsy could further direct some patients to alternative treatment strategies.

JMcD and JBOM would like to thank SULSA for funding. DSP thanks the University of Strathclyde for support through its Strategic Appointment and Investment Scheme. We compare a range of computational methods for the prediction of sublimation thermodynamics (enthalpy, entropy and free energy of sublimation). These include a model from theoretical chemistry that utilizes crystal lattice energy minimization (with the DMACRYS program) and QSPR models generated by both machine learning (Random Forest and Support Vector Machines) and regression (Partial Least Squares) methods. Using these methods we investigate the predictability of the enthalpy, entropy and free energy of sublimation, with consideration of whether such a method may be able to improve solubility prediction schemes. Previous work has suggested that the major source of error in solubility prediction schemes involving a thermodynamic cycle via the solid state is in the modeling of the free energy change away from the solid state. Yet contrary to this conclusion other work has found that the inclusion of terms such as the enthalpy of sublimation in QSPR methods does not improve the predictions of solubility. We suggest the use of theoretical chemistry terms, detailed explicitly in the methods section, as descriptors for the prediction of the enthalpy and free energy of sublimation. A dataset of 158 molecules with experimental sublimation thermodynamics values and some CSD refcodes has been collected from the literature and is provided with their original source references. Postprint

Pre-processing is an essential step in the analysis of spectral data. Mid-IR spectroscopy of biological samples is often subject to instrumental and sample specific variances which may often conceal valuable biological information. Whilst pre-processing can effectively reduce this unwanted variance, the plethora of possible processing steps has resulted in a lack of consensus in the field, often meaning that analysis outputs are not comparable. As pre-processing is specific to the sample under investigation, here we present a systematic approach for defining the optimum pre-processing protocol for biofluid ATR-FTIR spectroscopy. Using a trial-and-error based approach and a clinically relevant dataset describing control and brain cancer patients (3,897 spectra), the effects of pre-processing permutations on subsequent classification algorithms were observed, by assessing key diagnostic performance parameters, including sensitivity and specificity. It was found that optimum diagnostic performance correlated with the use of minimal binning and baseline correction, with derivative functions improving diagnostic performance most significantly. If smoothing is required, a Sovitzky-Golay approach was the preferred option in this investigation. Heavy binning appeared to reduce classification most significantly, alongside wavelet noise reduction (filter length ≥ 6), resulting in the lowest diagnostic performances of all pre-processing permutations tested.

In this work, we present a new method for predicting complex physicalchemical properties of organic molecules. The approach utilizes 3D convolutional neural network (ActivNet4) that uses solvent spatial distributions around solutes as input. These spatial distributions are obtained by a molecular theory called threedimensional reference interaction site model (3D-RISM). We have shown that the method allows one to achieve a good accuracy of prediction of bioconcentration factor (BCF) which is difficult to predict by direct application of methods of molecular theory or simulations. Our research demonstrates that combination of molecular theories with modern machine learning approaches can be effectively used for predicting properties that are otherwise inaccessible to purely theory-based models.

Objectives To determine the potential costs and health benefits of a serum-based spectroscopic triage tool for brain tumours, which could be developed to reduce diagnostic delays in the current clinical pathway. Design A model-based health pre-trial economic assessment. Decision tree models were constructed based on simplified diagnostic pathways. Models were populated with parameters identified from rapid reviews of the literature and clinical expert opinion. Setting Explored as a test in both primary and secondary care (neuroimaging) in the UK health service, as well as application to the USA. Participants Calculations based on an initial cohort of 10 000 patients. In primary care, it is estimated that the volume of tests would approach 75 000 per annum. The volume of tests in secondary care is estimated at 53 000 per annum. Main outcome measures The primary outcome measure was quality-adjusted life-years (QALY), which were employed to derive incremental cost-effectiveness ratios (ICER) in a cost-effectiveness analysis. Results Results indicate that using a blood-based spectroscopic test in both scenarios has the potential to be highly cost-effective in a health technology assessment agency decision-making process, as ICERs were well below standard threshold values of £20 000–£30 000 per QALY. This test may be cost-effective in both scenarios with test sensitivities and specificities as low as 80%; however, the price of the test would need to be lower (less than approximately £40). Conclusion Use of this test as triage tool in primary care has the potential to be both more effective and cost saving for the health service. In secondary care, this test would also be deemed more effective than the current diagnostic pathway.

The complex patterns observed from evaporated liquid drops have been examined extensively over the last 20 years. Complete understanding of drop deposition is vital in many medical processes, and one which is essential to the translation of biofluid spectroscopic disease diagnostics. The promising use of spectroscopy in disease diagnosis has been hindered by the complicated patterns left by dried biological fluids which may inhibit the clinical translation of this technology. Coffee-ring formation, cracking and gelation patterns have all been observed in biofluid drops, and with surface homogeneity being a key element to many spectroscopic techniques, experimental issues have been found to arise. A better understanding of the fundamental processes involved in a drying droplet could allow efficient progression in this research field, and ultimately benefit the population with the development of a reliable cancer diagnostic.

PRFFECT is a computer program to aid with spectral preprocessing and the development of classification models. Via a simple text interface, PRFFECT allows users to select wavenumber ranges, perform spectral preprocessing, carry out data partitioning (into training and testing datasets), run a Random Forest classification, compute statistical results, and identify important descriptors for the classification. The preprocessing options provided fall into four categories: binning, smoothing, normalisation, and baseline correction. The program outputs a wide-variety of useful data, including classification metrics and graphs showing the importance of individual wavenumbers to the classification models. As proof-of-concept, PRFFECT has been benchmarked on preprocessing and classification of four food analysis datasets. Sensitivities and specificities above 0.92 were obtained in all cases. The results show that different preprocessing procedures are optimal for different datasets. The PRFFECT software is available freely to the community via GitHub. Link: https://github.com/Palmer-Lab/PRFFECT .

The integral equation theory (IET) of molecular liquids has been an active area of academic research in theoretical and computational physical chemistry for over 40 years because it provides a consistent theoretical framework to describe the structural and thermodynamic properties of liquid-phase solutions. The theory can describe pure and mixed solvent systems (including anisotropic and nonequilibrium systems) and has already been used for theoretical studies of a vast range of problems in chemical physics / physical chemistry, molecular biology, colloids, soft matter, and electrochemistry. A consider- able advantage of IET is that it can be used to study speci fi c solute − solvent interactions, unlike continuum solvent models, but yet it requires considerably less computational expense than explicit solvent simulations.

D.S.P. is grateful for funding from the European Commission through a Marie Curie Intra-European Fellowship within the seventh European Community Framework Programme (FP7-PEOPLE-2010-IEF). D.S.P. thanks the University of Strathclyde for support through its Strategic Appointment and Investment Scheme. Computations were performed at the EPSRC funded ARCHIE-WeSt High Performance Computer (www.archie-west.ac.uk, EPSRC grant no. EP K0005861). J.B.O.M. thanks the Scottish Universities Life Sciences Alliance (SULSA) for financial support and EaStCHEM for access to the EaStCHEM Research Computing Facility. We report the results of testing quantitative structure-property relationships (QSPR) that were trained upon the same druglike molecules but two different sets of solubility data: (i) data extracted from several different sources from the published literature, for which the experimental uncertainty is estimated to be 0.6-0.7 log S units (referred to mol/L); (ii) data measured by a single accurate experimental method (CheqSol), for which experimental uncertainty is typically

Motivation Water molecules in protein binding sites play essential roles in biological processes. The popular 3D-RISM prediction method can calculate the solvent density distribution within minutes, but is difficult to convert it into explicit water molecules. Results We present GAsol, a tool that is capable of finding the network of water molecules that best fits a particular 3D-RISM density distribution in a fast and accurate manner and that outperforms other available tools by finding the globally optimal solution thanks to its genetic algorithm. Availability and implementation https://github.com/accsc/GAsol. BSD 3-clauses license Supplementary information Supplementary data are available at Bioinformatics online.

We demonstrate that using a pressure corrected three-dimensional reference interaction site model one can accurately predict salting-out (Setschenow's) constants for a wide range of organic compounds in aqueous solutions of NaCl. The approach, based on classical molecular force fields, offers an alternative to more heavily parametrized methods. Published by AIP Publishing.

We present a new approach for predicting solvation free energies in nonaqueous solvents. Utilizing the corresponding states principle, we estimate solvent Lennard-Jones parameters directly from their critical points. Combined with atomic solutes and the pressure corrected three-dimensional reference interaction site model (3D-RISM/PC+), the model gives accurate predictions for a wide range of nonpolar solvents, including olive oil. The results, obtained without electrostatic interactions and with a very coarse-grained solvent, provide an interesting alternative to widely used and heavily parametrized models.

Fourier transform infrared (FTIR) spectroscopy has long been established as an analytical technique for the measurement of vibrational modes of molecular systems. More recently, FTIR has been used for the analysis of biofluids with the aim of becoming a tool to aid diagnosis. For the clinician, this represents a convenient, fast, non-subjective option for the study of biofluids and the diagnosis of disease states. The patient also benefits from this method, as the procedure for the collection of serum is much less invasive and stressful than traditional biopsy. This is especially true of patients in whom brain cancer is suspected. A brain biopsy is very unpleasant for the patient, potentially dangerous and can occasionally be inconclusive. We therefore present a method for the diagnosis of brain cancer from serum samples using FTIR and machine learning techniques. The scope of the study involved 433 patients from whom were collected 9 spectra each in the range 600-4000 cm(-1). To begin the development of the novel method, various pre-processing steps were investigated and ranked in terms of final accuracy of the diagnosis. Random forest machine learning was utilised as a classifier to separate patients into cancer or non-cancer categories based upon the intensities of wavenumbers present in their spectra. Generalised 2D correlational analysis was then employed to further augment the machine learning, and also to establish spectral features important for the distinction between cancer and non-cancer serum samples. Using these methods, sensitivities of up to 92.8% and specificities of up to 91.5% were possible. Furthermore, ratiometrics were also investigated in order to establish any correlations present in the dataset. We show a rapid, computationally light, accurate, statistically robust methodology for the identification of spectral features present in differing disease states. With current advances in IR technology, such as the development of rapid discrete frequency collection, this approach is of importance to enable future clinical translation and enables IR to achieve its potential.

We present a theoretical/computational framework for accurate calculation of hydration free energies of ionized molecular species. The method is based on a molecular theory, 3D-RISM, combined with a recently developed pressure correction (PC+). The 3D-RISM/PC+ model can provide ∼3 kcal/mol hydration free energy accuracy for a large variety of ionic compounds, provided that the Galvani potential of water is taken into account. The results are compared with direct atomistic simulations. Several methodological aspects of hydration free energy calculations for charged species are discussed.

We demonstrate that the intrinsic aqueous solubility of crystalline druglike molecules can be estimated with reasonable accuracy from sublimation free energies calculated using crystal lattice simulations and hydration free energies calculated using the 3D Reference Interaction Site Model (3D-RISM) of the Integral Equation Theory of Molecular Liquids (IET). The solubilities of 25 crystalline druglike molecules taken from different chemical classes are predicted by the model with a correlation coefficient of R = 0.85 and a root mean square error (RMSE) equal to 1.45 log(10) S units, which is significantly more accurate than results obtained using implicit continuum solvent models. The method is not directly parametrized against experimental solubility data, and it offers a full computational characterization of the thermodynamics of transfer of the drug molecule from crystal phase to gas phase to dilute aqueous solution. Postprint

The Integral Equation Theory (IET) of Molecular Liquids is a theoretical framework for modelling solution phase behaviour that has recently found new applications in computational drug design. IET allows calculation of solvation thermodynamic parameters at significantly lower computational expense than explicit solvent simulations, but also provides information about the microscopic solvent structure that is not accessible by implicit continuum models. In this review we focus on recent advances in two fields of research using these methods: (i) calculation of the hydration free energies of bioactive molecules; (ii) modelling the aggregation of biomimetic molecules. In addition, we discuss sources of experimental solvation data for druglike molecules.

We report a method to dramatically improve the accuracy of hydration free energies (HFE) calculated by the 1D and 3D reference interaction site models (RISM) of molecular integral equation theory. It is shown that the errors in HFEs calculated by RISM approaches using the Gaussian fluctuations (GF) free energy functional are not random, but can be decomposed into linear combination of contributions from different structural elements of molecules (number of double bonds, number of OH groups, etc.). Therefore, by combining RISM/GF with cheminformatics, one can develop an accurate method for HFE prediction. We call this approach the structural description correction model (SDC) ( Ratkova et al. J. Phys. Chem. B 2010 , 114 , 12068 ). In this work, we investigated the prediction quality of the SDC model combined with 1D and 3D RISM approaches. In parallel, we analyzed the computational performance of these two methods. The SDC model parameters were obtained by fitting against a training set of 53 simple organic molecules. To demonstrate that the values of these parameters were transferable between different classes of molecules, the models were tested against 98 more complex molecules (including 38 polyfragment compounds). The results show that the 3D RISM/SDC model predicts the HFEs with very good accuracy (RMSE of 0.47 kcal/mol), while the 1D RISM approach provides only moderate accuracy (RMSE of 1.96 kcal/mol). However, a single 1D RISM/SDC calculation takes only a few seconds on a PC, whereas a single 3D RISM/SDC HFE calculation is approximately 100 times more computationally expensive. Therefore, we suggest that one should use the 1D RISM/SDC model for large-scale high-throughput screening of molecular hydration properties, while further refinement of these properties for selected compounds should be carried out with the more computationally expensive but more accurate 3D RISM/SDC model.

It is shown that a principal component analysis of a large set of internal coordinates can be used to define a nonredundant set of delocalized internal coordinates suitable for the calculation of harmonic vibrational normal modes. The selection of internal coordinates and the principal component analysis provide large degrees of freedom in extracting a nonredundant set of coordinates, and thus influence how the vibrational normal modes are described. It is shown that long-range coordinates may be especially suitable for describing low-frequency global deformation modes in proteins.

This paper attempts to elucidate differences in QSPR models of aqueous solubility (Log S), melting point (Tm), and octanol-water partition coefficient (Log P), three properties of pharmaceutical interest. For all three properties, Support Vector Machine models using 2D and 3D descriptors calculated in the Molecular Operating Environment were the best models. Octanol-water partition coefficient was the easiest property to predict, as indicated by the RMSE of the external test set and the coefficient of determination (RMSE = 0.73, r2 = 0.87). Melting point prediction, on the other hand, was the most difficult (RMSE = 52.8 degrees C, r2 = 0.46), and Log S statistics were intermediate between melting point and Log P prediction (RMSE = 0.900, r2 = 0.79). The data imply that for all three properties the lack of measured values at the extremes is a significant source of error. This source, however, does not entirely explain the poor melting point prediction, and we suggest that deficiencies in descriptors used in melting point prediction contribute significantly to the prediction errors.