Insights into performance evaluation of compound-protein interaction prediction methods

MOTIVATION

Machine-learning-based prediction of compound-protein interactions (CPIs) is important for drug design, screening and repurposing. Despite numerous recent publication with increasing methodological sophistication claiming consistent improvements in predictive accuracy, we have observed a number of fundamental issues in experiment design that produce overoptimistic estimates of model performance.

RESULTS

We systematically analyze the impact of several factors affecting generalization performance of CPI predictors that are overlooked in existing work: (i) similarity between training and test examples in cross-validation; (ii) synthesizing negative examples in absence of experimentally verified negative examples and (iii) alignment of evaluation protocol and performance metrics with real-world use of CPI predictors in screening large compound libraries. Using both state-of-the-art approaches by other researchers as well as a simple kernel-based baseline, we have found that effective assessment of generalization performance of CPI predictors requires careful control over similarity between training and test examples. We show that, under stringent performance assessment protocols, a simple kernel-based approach can exceed the predictive performance of existing state-of-the-art methods. We also show that random pairing for generating synthetic negative examples for training and performance evaluation results in models with better generalization in comparison to more sophisticated strategies used in existing studies. Our analyses indicate that using proposed experiment design strategies can offer significant improvements for CPI prediction leading to effective target compound screening for drug repurposing and discovery of putative chemical ligands of SARS-CoV-2-Spike and Human-ACE2 proteins.

AVAILABILITY AND IMPLEMENTATION

Code and supplementary material available at https://github.com/adibayaseen/HKRCPI.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Determination of hormoligosis of organophosphate insecticides against Phenacoccus solenopsis

Abstract Cotton mealybug is a highly invasive pest of agricultural crops worldwide. Major agriculturists most rely on the use of insecticides for the control of pesticides. So, the indiscriminate use of insecticides leads to resistance development in recent years. For this purpose, an experiment was conducted using different concentrations of the three insecticides (profenfos chlorpyrifos and triazophos) to check the hormoligosis effects against cotton mealybug (CMB) in laboratory conditions. Investigation of variations for % mortality of adults of CMB after three days revealed that all treatments had statistically significant (P ˂ 0.05). The highest mortality was observed at the highest concentrations of profenofos 2.4% (38.55%). After 7 days, all the treatments were significant with difference in means (P ˂ 0.05). The highest mortality was recorded at the highest dilution of pesticide profenofos 2.4% (77.11%). The values of fecundity and longevity exposed a valid difference among treatments (P ˂ 0.05). Maximum fecundity was observed at the concentration 2.4% (181.41%) and longevity showed (38.46%). The highest mortality was observed at a concentration of triazophos 4% (27.98%). For chlorpyriphos the highest mortality was examined at concentration 4% (24.79%). The fecundity showed a statistically significant difference for different concentrations of triazophos and chlorpyriphos (P ˂ 0.05). The results of the recent study provide valuable information regarding the selection of insecticides and hormoligosis effects. The study can be helpful in the implications of integrated pest management of P. solenopsis.

HemoNet: Predicting hemolytic activity of peptides with integrated feature learning

Quantifying the hemolytic activity of peptides is a crucial step in the discovery of novel therapeutic peptides. Computational methods are attractive in this domain due to their ability to guide wet-lab experimental discovery or screening of peptides based on their hemolytic activity. However, existing methods are unable to accurately model various important aspects of this predictive problem such as the role of N/C-terminal modifications, D- and L- amino acids, etc. In this work, we have developed a novel neural network-based approach called HemoNet for predicting the hemolytic activity of peptides. The proposed method captures the contextual importance of different amino acids in a given peptide sequence using a specialized feature embedding in conjunction with SMILES-based fingerprint representation of N/C-terminal modifications. We have analyzed the predictive performance of the proposed method using stratified cross-validation in comparison with previous methods, non-redundant cross-validation as well as validation on external peptides and clinical antimicrobial peptides. Our analysis shows the proposed approach achieves significantly better predictive performance (AUC-ROC of 88%) in comparison to previous approaches (HemoPI and HemoPred with AUC-ROC of 73%). HemoNet can be a useful tool in the search for novel therapeutic peptides. The python implementation of the proposed method is available at the URL: https://github.com/adibayaseen/HemoNet.

ISLAND: in-silico proteins binding affinity prediction using sequence information

BACKGROUND

Determining binding affinity in protein-protein interactions is important in the discovery and design of novel therapeutics and mutagenesis studies. Determination of binding affinity of proteins in the formation of protein complexes requires sophisticated, expensive and time-consuming experimentation which can be replaced with computational methods. Most computational prediction techniques require protein structures that limit their applicability to protein complexes with known structures. In this work, we explore sequence-based protein binding affinity prediction using machine learning.

METHOD

We have used protein sequence information instead of protein structures along with machine learning techniques to accurately predict the protein binding affinity.

RESULTS

We present our findings that the true generalization performance of even the state-of-the-art sequence-only predictor is far from satisfactory and that the development of machine learning methods for binding affinity prediction with improved generalization performance is still an open problem. We have also proposed a sequence-based novel protein binding affinity predictor called ISLAND which gives better accuracy than existing methods over the same validation set as well as on external independent test dataset. A cloud-based webserver implementation of ISLAND and its python code are available at https://sites.google.com/view/wajidarshad/software .

CONCLUSION

This paper highlights the fact that the true generalization performance of even the state-of-the-art sequence-only predictor of binding affinity is far from satisfactory and that the development of effective and practical methods in this domain is still an open problem.

Prevalence of Helicobacter pylori, gastric myoelectrical activity, gastric mucosal changes and dyspeptic symptoms before and after laparoscopic cholecystectomy

BACKGROUND/AIMS

Cholecystectomy may lead to anatomic and functional alterations which eventually induce reflux of duodenal contents with its sequlae. The aim of this study is to evaluate the prevalence of Helicobacter pylori (H. pylori), gastric myoelectrical activities and gastric mucosal changes before and after laparoscopic cholecystectomy.

METHODOLOGY

This prospective study has been carried out on 46 patients (20 M & 26 F) with mean age 41.7+/-0.2 years for whom laparoscopic cholecystectomy for gallstones was carried out. Prior to the operation and 1 year after, all patients were subjected to clinical assessment, upper gastrointestinal endoscopy, histopathology of antral mucosa, reflux gastritis score, detection of H. pylori and electrogastrography.

RESULTS

There was an increase in the postoperative suggestive symptoms of reflux gastritis compared to the preoperative: epigastric pain increased from 8 (17.4%) to 11 (23.39%) patients, nausea increased from 6 (13%) to 12 (26.1%) and bilious vomiting increased from 3 (6.5%) to 11 (23.9%) patients. Mild antral gastritis was detected endoscopically before laparoscopic cholecystectomy in 20 patients (43.5%) and increased to 27 patients (58.7%) after surgery. Meanwhile, severe antral gastritis and erosions were only detected after the operation in 10 (21.7%) patients, respectively. The histological results showed an increase of the histopathologic score of reflux gastritis after cholecystectomy from 4.28 (+/-1.56) to 9.28 (+/-1.99) (p<0.001). Active chronic superficial gastritis decreased from 23 (50%) to 13 (28.2%) patients while the inactive form increased from 15 (32.6%) to 23 (50%) patients. Also, chronic atrophic gastritis, intestinal metaplasia and dysplasia were detected postoperatively in 4 (8.6%) patients. The incidence of H. pylori infection was decreased from 32 (69.6%) to 19 (41.3%) patients (p<0.0001). Electrogastrography abnormal frequency decreased in fasting from 26.1% to 8.7% (p<0.001), and postprandial from 16.9% to 4.4% recording (p<0.002). On the other hand, there was an increase in the number of patients with decreased electrogastrography amplitude after a meal from 4.3% to 28.3% (p<0.0001).

CONCLUSIONS

Our study shows that dyspeptic symptoms, endoscopic and histologic gastric changes as well as electrogastrography abnormalities are present before and increase after cholecystectomy; meanwhile H. pylori colonization in gastric mucosa is decreased after cholecystectomy.

Does cholecystectomy affect antral mucosa? Endoscopic, histopathologic and DNA flow cytometric study

BACKGROUND/AIMS

Although cholecystectomy is still the "gold standard" for treatment of gallstones, this operation may be followed by gastric disorders. The aim of this study is to detect the effects of cholecystectomy on gastric antral mucosa.

METHODOLOGY

This prospective study has been carried out on 46 patients (20 M & 26 F) with mean age 41.7 +/- 0.2 years for whom simple cholecystectomy for gallstones was decided. Prior to the operation and 1 year after, patients were subjected to the following: clinical assessment, upper gastrointestinal endoscopy, histopathology of antral mucosa, detection of H. pylori and DNA flow cytometry.

RESULTS

There was an increase in the number of patients presenting suggestive symptoms of reflux gastritis: patients experiencing epigastric pain increased from 8 (17.4%) to 11 (23.39%) patients, nausea increased from 6 (13%) to 12 (26.1%) patients and bilious vomiting increased from 3 (6.5%) to 11 (23.9%) patients. Mild antral gastritis increased from 20 (43.5%) to 27 (58.7%) patients. Antral gastritis and antral erosions were detected only after the operation in 8 (17.4%) and 2 (4.3%) patients, respectively. The incidence of active chronic superficial gastritis decreased from 23 (50%) to 13 (28.2%) patients while the inactive form increased from 15 (32.6%) to 23 (50%) patients. Chronic atrophic gastritis, intestinal metaplasia and dysplasia were only detected postoperatively in 2 (4.3%) patients each. There was a decrease in the incidence of H. pylori infection from 32 (69.6) to 19 (41.3%) patients. DNA aneuploid pattern increased from 1 (2.2%) to 4 (8.7%) patients and there was a significant increase of DNA index from 1.01 (+/- 0.03) to 1.03 (+/- 0.05) (P < 0.005).

CONCLUSIONS

Changes in clinical, endoscopic and histopathologic findings suggest that cholecystectomy may affect gastric antral mucosa due to duodenogastric reflux. Flow cytometry may be used as an objective method for detection and evaluation of postcholecystectomy reflux gastritis.

Effect of dynorphin on insulin and somatostatin secretion, calcium uptake, and c-AMP levels in isolated rat islets of Langerhans

Dynorphin-[1-13], at concentrations of 5.8 X 10(-12) to 5.8 X 10(-9) M, stimulated insulin secretion from isolated islets of Langerhans of the rat, in medium containing 6 mM glucose. Higher concentrations of dynorphin had no significant effect on secretion. Dynorphin (5.8 X 10(-9) M) was unable to initiate insulin release from islets in the presence of 2 mM glucose, or to increase insulin secretion further in the presence of 20 mM glucose or 6 and 12 mM glyceraldehyde. Dynorphin-induced insulin secretion from islets was blocked by verapamil (5 microM) or by chlorpropamide (72 microM), but not by a mu opiate receptor antagonist, naloxone (0.11 microM), or by ICI 154129, a specific antagonist for the delta receptor (0.25 microM). Dynorphin had no effect on islet somatostatin secretion, under conditions in which insulin secretion was greatly stimulated. Glucose (20 mM) and glyceraldehyde (6 and 12 mM) significantly increased both insulin and somatostatin secretion. Dynorphin (5.8 X 10(-9) M) increased 45Ca2+ uptake into islets, and also increased intracellular islet c-AMP levels. These changes persisted when higher concentrations of dynorphin were used. These results suggest that (1) dynorphin is a very potent stimulus for insulin secretion; (2) dynorphin does not affect somatostatin secretion in static incubations of islets, in the same way as does glucose and glyceraldehyde; (3) dynorphin's effects may involve increased calcium ion movement and can be blocked by verapamil; (4) dynorphin can also increase islet c-AMP, and could thereby modulate the responsiveness of other secretagogues; (5) the actions of dynorphin on insulin secretion are not mediated by delta or mu opiate receptors in islets.