Dose-dependent transcriptional effects of lithium and adverse effect burden in a psychiatric cohort.

Lithium is the first-line treatment for bipolar disorder (BD), but there is a large variation in response rate and adverse effects. Although the molecular effects of lithium have been studied extensively, the specific mechanisms of action remain unclear. In particular, the molecular changes underlying lithium adverse effects are little known. Multiple linear regression analyses of lithium serum concentrations and global gene expression levels in whole blood were carried out using a large case-control sample (n = 1450). Self-reported adverse effects of lithium were assessed with the "Udvalg for Kliniske Undersøgelser" (UKU) adverse effect rating scale, and regression analysis was used to identify significant associations between lithium-related genes and six of the most common adverse effects. Serum concentrations of lithium were significantly associated with the expression levels of 52 genes (FDR < 0.01), largely replicating previous results. We found 32 up-regulated genes and 20 down-regulated genes in lithium users compared to non-users. The down-regulated gene set was enriched for several processes related to the translational machinery. Two adverse effects were significantly associated (p < 0.01) with three or more lithium-associated genes: tremor (FAM13A-AS1, FAR2, ITGAX, RWDD1, and STARD10) and xerostomia (ANKRD13A, FAR2, RPS8, and RWDD1). The adverse effect association with the largest effect was between CAMK1D expression and nausea/vomiting. These results suggest putative transcriptional mechanisms that may predict lithium adverse effects, and could thus have a large potential for informing clinical practice.

Genetic loci shared between major depression and intelligence with mixed directions of effect.

Genome-wide association studies (GWAS) have identified several common genetic variants influencing major depression and general cognitive abilities, but little is known about whether the two share any of their genetic aetiology. Here we investigate shared genomic architectures between major depression (MD) and general intelligence (INT) with the MiXeR statistical tool and their overlapping susceptibility loci with conjunctional false discovery rate (conjFDR), which evaluate the level of overlap in genetic variants and improve the power for gene discovery between two phenotypes. We analysed GWAS data on MD (n = 480,359) and INT (n = 269,867) to characterize polygenic architecture and identify genetic loci shared between these phenotypes. Despite non-significant genetic correlation (rg = -0.0148, P = 0.50), we observed large polygenic overlap and identified 92 loci shared between MD and INT at conjFDR < 0.05. Among the shared loci, 69 and 64 are new for MD and INT, respectively. Our study demonstrates polygenic overlap between these phenotypes with a balanced mixture of effect.

Disconnected Maximum Common Substructures under Constraints.

The maximum common substructure (MCS) problem is an important, well-studied problem in cheminformatics. It is applied in several application scenarios like molecule superimposition and scaffold detection or as a similarity measure in virtual screening and clustering. In many cases, the connected MCS is preferred since it is faster to calculate and a highly fragmented MCS is not very meaningful from a chemical point of view. Nevertheless, a disconnected MCS (dMCS) can be very instructive if it consists of reasonably sized molecular parts connected by variable groups. We present a new algorithm named RIMACS, which is able to calculate the dMCS under constraints. We can control the maximum number of connected components and their minimal size using a modified local substructure mapping approach. A formal proof of correctness is provided as well as extended runtime evaluations on chemical data. The evaluation of RIMACS shows that a small number of connected components helps us to improve MCS similarity in a meaningful way while keeping the runtime requirements in a reasonable range.

Identification of genetic overlap and novel risk loci for attention-deficit/hyperactivity disorder and bipolar disorder.

Differential diagnosis between childhood onset attention-deficit/hyperactivity disorder (ADHD) and bipolar disorder (BD) remains a challenge, mainly due to overlapping symptoms and high rates of comorbidity. Despite this, genetic correlation reported for these disorders is low and non-significant. Here we aimed to better characterize the genetic architecture of these disorders utilizing recent large genome-wide association studies (GWAS). We analyzed independent GWAS summary statistics for ADHD (19,099 cases and 34,194 controls) and BD (20,352 cases and 31,358 controls) applying the conditional/conjunctional false discovery rate (condFDR/conjFDR) statistical framework that increases the power to detect novel phenotype-specific and shared loci by leveraging the combined power of two GWAS. We observed cross-trait polygenic enrichment for ADHD conditioned on associations with BD, and vice versa. Leveraging this enrichment, we identified 19 novel ADHD risk loci and 40 novel BD risk loci at condFDR <0.05. Further, we identified five loci jointly associated with ADHD and BD (conjFDR < 0.05). Interestingly, these five loci show concordant directions of effect for ADHD and BD. These results highlight a shared underlying genetic risk for ADHD and BD which may help to explain the high comorbidity rates and difficulties in differentiating between ADHD and BD in the clinic. Improving our understanding of the underlying genetic architecture of these disorders may aid in the development of novel stratification tools to help reduce these diagnostic difficulties.

Genome-wide Association Analysis of Parkinson's Disease and Schizophrenia Reveals Shared Genetic Architecture and Identifies Novel Risk Loci.

BACKGROUND: Parkinson's disease (PD) and schizophrenia (SCZ) are heritable brain disorders that involve dysregulation of the dopaminergic system. Epidemiological studies have reported potential comorbidity between the disorders, and movement disturbances are common in patients with SCZ before treatment with antipsychotic drugs. Despite this, little is known about shared genetic etiology between the disorders. METHODS: We analyzed recent large genome-wide association studies on patients with SCZ (N = 77,096) and PD (N = 417,508) using a conditional/conjunctional false discovery rate (FDR) approach to evaluate overlap in common genetic variants and improve statistical power for genetic discovery. Using a variety of biological resources, we functionally characterized the identified genomic loci. RESULTS: We observed genetic enrichment in PD conditional on associations with SCZ and vice versa, indicating polygenic overlap. We then leveraged this cross-trait enrichment using conditional FDR analysis and identified 9 novel PD risk loci and 1 novel SCZ locus at conditional FDR < .01. Furthermore, we identified 9 genomic loci jointly associated with PD and SCZ at conjunctional FDR < .05. There was an even distribution of antagonistic and agonistic effect directions among the shared loci, in line with the insignificant genetic correlation between the disorders. Of 67 genes mapped to the shared loci, 65 are expressed in the human brain and show cell type-specific expression profiles. CONCLUSIONS: Altogether, the study increases understanding of the genetic architectures underlying SCZ and PD, indicating that common molecular genetic mechanisms may contribute to overlapping pathophysiological and clinical features between the disorders.

Identification of Genetic Loci Shared Between Attention-Deficit/Hyperactivity Disorder, Intelligence, and Educational Attainment.

BACKGROUND: Attention-deficit/hyperactivity disorder (ADHD) is a neurodevelopmental disorder that is consistently associated with lower levels of educational attainment. A recent large genome-wide association study identified common gene variants associated with ADHD, but most of the genetic architecture remains unknown. METHODS: We analyzed independent genome-wide association study summary statistics for ADHD (19,099 cases and 34,194 controls), educational attainment (N = 842,499), and general intelligence (N = 269,867) using a conditional/conjunctional false discovery rate (FDR) statistical framework that increases power of discovery by conditioning the FDR on overlapping associations. The genetic variants identified were characterized in terms of function, expression, and biological processes. RESULTS: We identified 58 linkage disequilibrium-independent ADHD-associated loci (conditional FDR < 0.01), of which 30 were shared between ADHD and educational attainment or general intelligence (conjunctional FDR < 0.01) and 46 were novel risk loci for ADHD. CONCLUSIONS: These results expand on previous genetic and epidemiological studies and support the hypothesis of a shared genetic basis between these phenotypes. Although the clinical utility of the identified loci remains to be determined, they can be used as resources to guide future studies aiming to disentangle the complex etiologies of ADHD, educational attainment, and general intelligence.

Shared Genetic Loci Between Body Mass Index and Major Psychiatric Disorders: A Genome-wide Association Study.

Importance: People with major psychiatric disorders (MPDs) have a 10- to 20-year shorter life span than the rest of the population, and this difference is mainly due to comorbid cardiovascular diseases. Genome-wide association studies have identified common variants involved in schizophrenia (SCZ), bipolar disorder (BIP), and major depression (MD) and body mass index (BMI), a key cardiometabolic risk factor. However, genetic variants jointly influencing MPD and BMI remain largely unknown. Objective: To assess the extent of the overlap between the genetic architectures of MPDs and BMI and identify genetic loci shared between them. Design, Setting, and Participants: Using a conditional false discovery rate statistical framework, independent genome-wide association study data on individuals with SCZ (n = 82 315), BIP (n = 51 710), MD (n = 480 359), and BMI (n = 795 640) were analyzed. The UK Biobank cohort (n = 29 740) was excluded from the MD data set to avoid sample overlap. Data were collected from August 2017 to May 2018, and analysis began July 2018. Main Outcomes and Measures: The primary outcomes were a list of genetic loci shared between BMI and MPDs and their functional pathways. Results: Genome-wide association study data from 1 380 284 participants were analyzed, and the genetic correlation between BMI and MPDs varied (SCZ: r for genetic = -0.11, P = 2.1 × 10-10; BIP: r for genetic = -0.06, P = .0103; MD: r for genetic = 0.12, P = 6.7 × 10-10). Overall, 63, 17, and 32 loci shared between BMI and SCZ, BIP, and MD, respectively, were analyzed at conjunctional false discovery rate less than 0.01. Of the shared loci, 34% (73 of 213) in SCZ, 52% (36 of 69) in BIP, and 57% (56 of 99) in MD had risk alleles associated with higher BMI (conjunctional false discovery rate <0.05), while the rest had opposite directions of associations. Functional analyses indicated that the overlapping loci are involved in several pathways including neurodevelopment, neurotransmitter signaling, and intracellular processes, and the loci with concordant and opposite association directions pointed mostly to different pathways. Conclusions and Relevance: In this genome-wide association study, extensive polygenic overlap between BMI and SCZ, BIP, and MD were found, and 111 shared genetic loci were identified, implicating novel functional mechanisms. There was mixture of association directions in SCZ and BMI, albeit with a preponderance of discordant ones.

Correction: Genome-wide analysis reveals extensive genetic overlap between schizophrenia, bipolar disorder, and intelligence.

A correction to this paper has been published and can be accessed via a link at the top of the paper.

Genome-wide analysis reveals extensive genetic overlap between schizophrenia, bipolar disorder, and intelligence.

Schizophrenia (SCZ) and bipolar disorder (BD) are severe mental disorders associated with cognitive impairment, which is considered a major determinant of functional outcome. Despite this, the etiology of the cognitive impairment is poorly understood, and no satisfactory cognitive treatments exist. Increasing evidence indicates that genetic risk for SCZ may contribute to cognitive impairment, whereas the genetic relationship between BD and cognitive function remains unclear. Here, we combined large genome-wide association study data on SCZ (n = 82,315), BD (n = 51,710), and general intelligence (n = 269,867) to investigate overlap in common genetic variants using conditional false discovery rate (condFDR) analysis. We observed substantial genetic enrichment in both SCZ and BD conditional on associations with intelligence indicating polygenic overlap. Using condFDR analysis, we leveraged this enrichment to increase statistical power and identified 75 distinct genomic loci associated with both SCZ and intelligence, and 12 loci associated with both BD and intelligence at conjunctional FDR < 0.01. Among these loci, 20 are novel for SCZ, and four are novel for BD. Most SCZ risk alleles (61 of 75, 81%) were associated with poorer cognitive performance, whereas most BD risk alleles (9 of 12, 75%) were associated with better cognitive performance. A gene set analysis of the loci shared between SCZ and intelligence implicated biological processes related to neurodevelopment, synaptic integrity, and neurotransmission; the same analysis for BD was underpowered. Altogether, the study demonstrates that both SCZ and BD share genetic influences with intelligence, albeit in a different manner, providing new insights into their genetic architectures.

Alterations in Schizophrenia-Associated Genes Can Lead to Increased Power in Delta Oscillations.

Genome-wide association studies have implicated many ion channels in schizophrenia pathophysiology. Although the functions of these channels are relatively well characterized by single-cell studies, the contributions of common variation in these channels to neurophysiological biomarkers and symptoms of schizophrenia remain elusive. Here, using computational modeling, we show that a common biomarker of schizophrenia, namely, an increase in delta-oscillation power, may be a direct consequence of altered expression or kinetics of voltage-gated ion channels or calcium transporters. Our model of a circuit of layer V pyramidal cells highlights multiple types of schizophrenia-related variants that contribute to altered dynamics in the delta-frequency band. Moreover, our model predicts that the same membrane mechanisms that increase the layer V pyramidal cell network gain and response to delta-frequency oscillations may also cause a deficit in a single-cell correlate of the prepulse inhibition, which is a behavioral biomarker highly associated with schizophrenia.

Exploring the Wnt signaling pathway in schizophrenia and bipolar disorder.

The Wnt signaling pathway plays a crucial role in neurodevelopment and in regulating the function and structure of the adult nervous system. Schizophrenia (SCZ) and bipolar disorder (BD) are severe mental disorders with evidence of subtle neurodevelopmental, structural and functional neuronal abnormalities. We aimed to elucidate the role of aberrant regulation of the Wnt system in these disorders by evaluating plasma levels of secreted Wnt modulators in patients (SCZ = 551 and BD = 246) and healthy controls (HCs = 639) using enzyme immune-assay. We also investigated the expression of 141 Wnt-related genes in whole blood in a subsample (SCZ = 338, BD = 241, and HCs = 263) using microarray analysis. Both SCZ and BD had dysregulated mRNA expression of Wnt-related genes favoring attenuated canonical (beta-catenin-dependent) signaling, and there were also indices of enhanced non-canonical Wnt signaling. In particular, FZD7, which may activate all Wnt pathways, but favors non-canonical signaling, and NFATc3, a downstream transcription factor and readout of the non-canonical Wnt/Ca2+ pathway, were significantly increased in SCZ and BD (p < 3 × 10-4). Furthermore, patients had lower plasma levels of soluble dickkopf 1 and sclerostin (p < 0.01) compared with HC. Our findings suggest that SCZ and BD are characterized by abnormal Wnt gene expression and plasma protein levels, and we propose that drugs targeting the Wnt pathway may have a role in the treatment of severe mental disorders.

Attenuated Notch signaling in schizophrenia and bipolar disorder.

The Notch signaling pathway plays a crucial role in neurodevelopment and in adult brain homeostasis. We aimed to further investigate Notch pathway activity in bipolar disorder (BD) and schizophrenia (SCZ) by conducting a pathway analysis. We measured plasma levels of Notch ligands (DLL1 and DLK1) using enzyme immunoassays in a large sample of patients (SCZ n = 551, BD n = 246) and healthy controls (HC n = 639). We also determined Notch pathway related gene expression levels by microarray analyses from whole blood in a subsample (SCZ n = 338, BD n = 241 and HC n = 263). We found significantly elevated Notch ligand levels in plasma in both SCZ and BD compared to HC. Significant gene expression findings included increased levels of RFNG and KAT2B (p < 0.001), and decreased levels of PSEN1 and CREBBP in both patient groups (p < 0.001). RBPJ was significantly lower in SCZ vs HC (p < 0.001), and patients using lithium had higher levels of RBPJ (p < 0.001). We provide evidence of altered Notch signaling in both SCZ and BD compared to HC, and suggest that Notch signaling pathway may be disturbed in these disorders. Lithium may ameliorate aberrant Notch signaling. We propose that drugs targeting Notch pathway could be relevant in the treatment of psychotic disorders.

Novel Loci Associated With Attention-Deficit/Hyperactivity Disorder Are Revealed by Leveraging Polygenic Overlap With Educational Attainment.

OBJECTIVE: Attention-deficit/hyperactivity disorder (ADHD) is a common and highly heritable psychiatric condition. By exploiting the reported relationship between ADHD and educational attainment (EA), we aimed to improve discovery of ADHD-associated genetic variants and to investigate genetic overlap between these phenotypes. METHOD: A conditional/conjunctional false discovery rate (condFDR/conjFDR) method was applied to genome-wide association study (GWAS) data on ADHD (2,064 trios, 896 cases, and 2,455 controls) and EA (n=328,917) to identify ADHD-associated loci and loci overlapping between ADHD and EA. Identified single nucleotide polymorphisms (SNPs) were tested for association in an independent population-based study of ADHD symptoms (n=17,666). Genetic correlation between ADHD and EA was estimated using LD score regression and Pearson correlation. RESULTS: At levels of condFDR<0.01 and conjFDR<0.05, we identified 5 ADHD-associated loci, 3 of these being shared between ADHD and EA. None of these loci had been identified in the primary ADHD GWAS, demonstrating the increased power provided by the condFDR/conjFDR analysis. Leading SNPs for 4 of 5 identified regions are in introns of protein coding genes (KDM4A, MEF2C, PINK1, RUNX1T1), whereas the remaining one is an intergenic SNP on chromosome 2 at 2p24. Consistent direction of effects in the independent study of ADHD symptoms was shown for 4 of 5 identified loci. A polygenic overlap between ADHD and EA was supported by significant genetic correlation (rg=-0.403, p=7.90×10-8) and >10-fold mutual enrichment of SNPs associated with both traits. CONCLUSION: We identified 5 novel loci associated with ADHD and provided evidence for a shared genetic basis between ADHD and EA. These findings could aid understanding of the genetic risk architecture of ADHD and its relation to EA.

Identification of Genetic Loci Jointly Influencing Schizophrenia Risk and the Cognitive Traits of Verbal-Numerical Reasoning, Reaction Time, and General Cognitive Function.

Importance: Schizophrenia is associated with widespread cognitive impairments. Although cognitive deficits are one of the factors most strongly associated with functional outcome in schizophrenia, current treatment strategies largely fail to ameliorate these impairments. To develop more efficient treatment strategies in patients with schizophrenia, a better understanding of the pathogenesis of these cognitive deficits is needed. Accumulating evidence indicates that genetic risk of schizophrenia may contribute to cognitive dysfunction. Objective: To identify genomic regions jointly influencing schizophrenia and the cognitive domains of reaction time and verbal-numerical reasoning, as well as general cognitive function, a phenotype that captures the shared variation in performance across cognitive domains. Design, Setting, and Participants: Combining data from genome-wide association studies from multiple phenotypes using conditional false discovery rate analysis provides increased power to discover genetic variants and could elucidate shared molecular genetic mechanisms. Data from the following genome-wide association studies, published from July 24, 2014, to January 17, 2017, were combined: schizophrenia in the Psychiatric Genomics Consortium cohort (n = 79 757 [cases, 34 486; controls, 45 271]); verbal-numerical reasoning (n = 36 035) and reaction time (n = 111 483) in the UK Biobank cohort; and general cognitive function in CHARGE (Cohorts for Heart and Aging Research in Genomic Epidemiology) (n = 53 949) and COGENT (Cognitive Genomics Consortium) (n = 27 888). Main Outcomes and Measures: Genetic loci identified by conditional false discovery rate analysis. Brain messenger RNA expression and brain expression quantitative trait locus functionality were determined. Results: Among the participants in the genome-wide association studies, 21 loci jointly influencing schizophrenia and cognitive traits were identified: 2 loci shared between schizophrenia and verbal-numerical reasoning, 6 loci shared between schizophrenia and reaction time, and 14 loci shared between schizophrenia and general cognitive function. One locus was shared between schizophrenia and 2 cognitive traits and represented the strongest shared signal detected (nearest gene TCF20; chromosome 22q13.2), and was shared between schizophrenia (z score, 5.01; P = 5.53 × 10-7), general cognitive function (z score, -4.43; P = 9.42 × 10-6), and verbal-numerical reasoning (z score, -5.43; P = 5.64 × 10-8). For 18 loci, schizophrenia risk alleles were associated with poorer cognitive performance. The implicated genes are expressed in the developmental and adult human brain. Replicable expression quantitative trait locus functionality was identified for 4 loci in the adult human brain. Conclusions and Relevance: The discovered loci improve the understanding of the common genetic basis underlying schizophrenia and cognitive function, suggesting novel molecular genetic mechanisms.

ProPairs: A Data Set for Protein-Protein Docking.

ProPairs is a data set of crystal structures of protein complexes defined as biological assemblies in the protein data bank (PDB), which are classified as legitimate protein-protein docking complexes by also identifying the corresponding unbound protein structures in the PDB. The underlying program selecting suitable protein complexes, also called ProPairs, is an automated method to extract structures of legitimate protein docking complexes and their unbound partner proteins from the PDB which fulfill specific criteria. In this way a total of 5,642 protein complexes have been identified with 11,600 different decompositions in unbound protein pairs yielding legitimate protein docking partners. After removing sequence redundancy (requiring a sequence identity of the residues in the interface of less than 40%), 2,070 different legitimate protein docking complexes remain. For 810 of these protein docking complexes, both docking partners possess corresponding unbound structures in the PDB. From the 2,070 nonredundant protein docking complexes there are 417 which possess a cofactor at the interface. From the 176 protein docking complexes of the Protein-Protein Docking Benchmark 4.0 (DB4.0) data set, 13 differ from the ProPairs data set. Twelve of them differ with respect to the composition of the unbound structures but are contained in the large redundant ProPairs data set. One protein docking complex of the DB4.0 data set is not contained in ProPairs since the biological assembly specified in the PDB is wrong (PDB id 1d6r ). For one protein complex (PDB id 1bgx ) the DB4.0 data set uses a fabricated unbound structure. For public use interactive online access is provided to the ProPairs data set of nonredundant protein docking complexes along with the source code of the underlying method [ http://propairs.github.io].

Optimized distance-dependent atom-pair-based potential DOOP for protein structure prediction.

The DOcking decoy-based Optimized Potential (DOOP) energy function for protein structure prediction is based on empirical distance-dependent atom-pair interactions. To optimize the atom-pair interactions, native protein structures are decomposed into polypeptide chain segments that correspond to structural motives involving complete secondary structure elements. They constitute near native ligand-receptor systems (or just pairs). Thus, a total of 8609 ligand-receptor systems were prepared from 954 selected proteins. For each of these hypothetical ligand-receptor systems, 1000 evenly sampled docking decoys with 0-10 Å interface root-mean-square-deviation (iRMSD) were generated with a method used before for protein-protein docking. A neural network-based optimization method was applied to derive the optimized energy parameters using these decoys so that the energy function mimics the funnel-like energy landscape for the interaction between these hypothetical ligand-receptor systems. Thus, our method hierarchically models the overall funnel-like energy landscape of native protein structures. The resulting energy function was tested on several commonly used decoy sets for native protein structure recognition and compared with other statistical potentials. In combination with a torsion potential term which describes the local conformational preference, the atom-pair-based potential outperforms other reported statistical energy functions in correct ranking of native protein structures for a variety of decoy sets. This is especially the case for the most challenging ROSETTA decoy set, although it does not take into account side chain orientation-dependence explicitly. The DOOP energy function for protein structure prediction, the underlying database of protein structures with hypothetical ligand-receptor systems and their decoys are freely available at http://agknapp.chemie.fu-berlin.de/doop/.

An adaptive self-healing ionic liquid nanocomposite membrane for olefin-paraffin separations.

An adaptive self-healing ionic liquid nanocomposite membrane comprising a multi-layer support structure hosting the ionic salt [Ag](+) [Tf(2) N](-) is used for the separation of the olefin propylene and the paraffin propane. The ionic salt renders liquid like upon complexation with propylene, resulting in facilitated transport of propylene over propane at benchmark-setting selectivity and permeance levels. The contacting with acetylene causes the ionic salt to liquefy without showing evidence of forming explosive silver acetylide.

Ionic liquid silver salt complexes for propene/propane separation.

Properties of the room-temperature liquid complex salt [Ag(propene)(x)][Tf(2)N] have been studied to probe its suitability for acting as active separation layer in immobilised liquid membrane (ILM) concepts for propane/propene separation. The pressure/temperature range of complex formation has been determined and the thermal properties of Ag[Tf(2)N] and [Ag(propene)(x)][Tf(2)N] have been studied by DSC (differential scanning calorimetry) and TGA (thermogravimetric analysis) measurements. Pressure dependent measurements of solubility and diffusivity showed that the observed membrane selectivity is dominated by the solubility selectivity. The self-diffusion coefficient of propene is always smaller compared to propane as propene is temporarily bound to the silver ion in the [Ag(propene)(x)][Tf(2)N] ionic liquid.

Predicting protein complex geometries with linear scoring functions.

Protein-Protein interactions play an important role in many cellular processes. However experimental determination of the protein complex structure is quite difficult and time consuming. Hence, there is need for fast and accurate in silico protein docking methods. These methods generally consist of two stages: (i) a sampling algorithm that generates a large number of candidate complex geometries (decoys), and (ii) a scoring function that ranks these decoys such that nearnative decoys are higher ranked than other decoys. We have recently developed a neural network based scoring function that performed better than other state-of-the-art scoring functions on a benchmark of 65 protein complexes. Here, we use similar ideas to develop a method that is based on linear scoring functions. We compare the linear scoring function of the present study with other knowledge-based scoring functions such as ZDOCK 3.0, ZRANK and the previously developed neural network. Despite its simplicity the linear scoring function performs as good as the compared state-of-the-art methods and predictions are simple and rapid to compute.

Predicting protein complex geometries with a neural network.

A major challenge of the protein docking problem is to define scoring functions that can distinguish near-native protein complex geometries from a large number of non-native geometries (decoys) generated with noncomplexed protein structures (unbound docking). In this study, we have constructed a neural network that employs the information from atom-pair distance distributions of a large number of decoys to predict protein complex geometries. We found that docking prediction can be significantly improved using two different types of polar hydrogen atoms. To train the neural network, 2000 near-native decoys of even distance distribution were used for each of the 185 considered protein complexes. The neural network normalizes the information from different protein complexes using an additional protein complex identity input neuron for each complex. The parameters of the neural network were determined such that they mimic a scoring funnel in the neighborhood of the native complex structure. The neural network approach avoids the reference state problem, which occurs in deriving knowledge-based energy functions for scoring. We show that a distance-dependent atom pair potential performs much better than a simple atom-pair contact potential. We have compared the performance of our scoring function with other empirical and knowledge-based scoring functions such as ZDOCK 3.0, ZRANK, ITScore-PP, EMPIRE, and RosettaDock. In spite of the simplicity of the method and its functional form, our neural network-based scoring function achieves a reasonable performance in rigid-body unbound docking of proteins. Proteins 2010. (c) 2009 Wiley-Liss, Inc.