ACE: adaptive cluster expansion for maximum entropy graphical model inference.

MOTIVATION: Graphical models are often employed to interpret patterns of correlations observed in data through a network of interactions between the variables. Recently, Ising/Potts models, also known as Markov random fields, have been productively applied to diverse problems in biology, including the prediction of structural contacts from protein sequence data and the description of neural activity patterns. However, inference of such models is a challenging computational problem that cannot be solved exactly. Here, we describe the adaptive cluster expansion (ACE) method to quickly and accurately infer Ising or Potts models based on correlation data. ACE avoids overfitting by constructing a sparse network of interactions sufficient to reproduce the observed correlation data within the statistical error expected due to finite sampling. When convergence of the ACE algorithm is slow, we combine it with a Boltzmann Machine Learning algorithm (BML). We illustrate this method on a variety of biological and artificial datasets and compare it to state-of-the-art approximate methods such as Gaussian and pseudo-likelihood inference. RESULTS: We show that ACE accurately reproduces the true parameters of the underlying model when they are known, and yields accurate statistical descriptions of both biological and artificial data. Models inferred by ACE more accurately describe the statistics of the data, including both the constrained low-order correlations and unconstrained higher-order correlations, compared to those obtained by faster Gaussian and pseudo-likelihood methods. These alternative approaches can recover the structure of the interaction network but typically not the correct strength of interactions, resulting in less accurate generative models. AVAILABILITY AND IMPLEMENTATION: The ACE source code, user manual and tutorials with the example data and filtered correlations described herein are freely available on GitHub at https://github.com/johnbarton/ACE CONTACTS: jpbarton@mit.edu, cocco@lps.ens.frSupplementary information: Supplementary data are available at Bioinformatics online.

Neural assemblies revealed by inferred connectivity-based models of prefrontal cortex recordings.

We present two graphical model-based approaches to analyse the distribution of neural activities in the prefrontal cortex of behaving rats. The first method aims at identifying cell assemblies, groups of synchronously activating neurons possibly representing the units of neural coding and memory. A graphical (Ising) model distribution of snapshots of the neural activities, with an effective connectivity matrix reproducing the correlation statistics, is inferred from multi-electrode recordings, and then simulated in the presence of a virtual external drive, favoring high activity (multi-neuron) configurations. As the drive increases groups of neurons may activate together, and reveal the existence of cell assemblies. The identified groups are then showed to strongly coactivate in the neural spiking data and to be highly specific of the inferred connectivity network, which offers a sparse representation of the correlation pattern across neural cells. The second method relies on the inference of a Generalized Linear Model, in which spiking events are integrated over time by neurons through an effective connectivity matrix. The functional connectivity matrices inferred with the two approaches are compared. Sampling of the inferred GLM distribution allows us to study the spatio-temporal patterns of activation of neurons within the identified cell assemblies, particularly their activation order: the prevalence of one order with respect to the others is weak and reflects the neuron average firing rates and the strength of the largest effective connections. Other properties of the identified cell assemblies (spatial distribution of coactivation events and firing rates of coactivating neurons) are discussed.

Stochastic ratchet mechanisms for replacement of proteins bound to DNA.

Experiments indicate that unbinding rates of proteins from DNA can depend on the concentration of proteins in nearby solution. Here we present a theory of multistep replacement of DNA-bound proteins by solution-phase proteins. For four different kinetic scenarios we calculate the dependence of protein unbinding and replacement rates on solution protein concentration. We find (1) strong effects of progressive "rezipping" of the solution-phase protein onto DNA sites liberated by "unzipping" of the originally bound protein, (2) that a model in which solution-phase proteins bind nonspecifically to DNA can describe experiments on exchanges between the nonspecific DNA-binding proteins Fis-Fis and Fis-HU, and (3) that a binding specific model describes experiments on the exchange of CueR proteins on specific binding sites.

Adaptive cluster expansion for inferring boltzmann machines with noisy data.

We introduce a procedure to infer the interactions among a set of binary variables, based on their sampled frequencies and pairwise correlations. The algorithm builds the clusters of variables contributing most to the entropy of the inferred Ising model and rejects the small contributions due to the sampling noise. Our procedure successfully recovers benchmark Ising models even at criticality and in the low temperature phase, and is applied to neurobiological data.

Data on soil physicochemical properties and chemical composition of rainfall and of throughfall and stemflow generated by Turkey oak trees (Quercus cerris L.) in acid and sub-alkaline soils.

We report data on the physicochemical properties of soils collected in two adjacent areas, one acid and one sub-alkaline, both developed on sequential beds of Plio-pleistocene marine sediments, and on the chemical composition of ecological solutions (rainfall, throughfall and stemflow) separately collected in the two areas. Throughfall and stemflow were generated by Turkey oak trees (Quercus cerris L.), which was the dominant tree species in both study areas. These data are related to the original article "Soil affects throughfall and stemflow under Turkey oak (Quercus cerris L.)" (Corti et al., 2019) [1].

Inferring DNA sequences from mechanical unzipping data: the large-bandwidth case.

The complementary strands of DNA molecules can be separated when stretched apart by a force; the unzipping signal is correlated to the base content of the sequence but is affected by thermal and instrumental noise. We consider here the ideal case where opening events are known to a very good time resolution (very large bandwidth), and study how the sequence can be reconstructed from the unzipping data. Our approach relies on the use of statistical Bayesian inference and of Viterbi decoding algorithm. Performances are studied numerically on Monte Carlo generated data, and analytically. We show how multiple unzippings of the same molecule may be exploited to improve the quality of the prediction, and calculate analytically the number of required unzippings as a function of the bandwidth, the sequence content, and the elasticity parameters of the unzipped strands.

[Intellectual quotient and environmental exposure to low doses of lead]. / Quoziente intellettivo ed esposizione ambientale a basse dosi di piombo.

The effects of low-level lead exposure on children's intelligence quotient (IQ) were investigated in 64 Sardinian adolescents (13-16 years old). To estimate potential early adverse effects on the Central Nervous System (CNS) due to very low-level lead, the Intelligence Quotient (IQ) of adolescents with present blood lead concentrations (PbB) generally below 10 microg/dl was measured. We analyzed blood lead concentration and individual IQ of 32 Sardinian children living in Portoscuso, a town 2 Km far from a lead smelter, and of other 32 controls living in S. Antioco, a town about 15 Km far from the same smelter. The Wechsler Intelligence Scale for Children Revised (WISC-R) was administered. The relation between IQ and blood lead concentration was estimated by linear multivariate models adjusting for several potential confounders, such as the educational and socio-economic level of the parents. The blood lead concentration was in average significantly higher in the Portoscuso group compared to controls. The linear model applied to the total population studied (n 64) showed that the blood lead concentration was inversely and significantly associated with IQ, with an extrapolated decline of 1.29 points in total IQ for each microg/dl increase of lead blood concentration. According to the recent scientific literature on this topic, results of our pilot study suggest the need to further lower the blood lead concentration for children to a threshold significantly below 10 microg/dl, value till now considered "safe" for the children's CNS.

Transglutaminase 2 ablation leads to mitophagy impairment associated with a metabolic shift towards aerobic glycolysis.

Macroautophagy selectively degrades dysfunctional mitochondria by a process known as mitophagy. Here we demonstrate the involvement of transglutaminase 2 (TG2) in the turnover and degradation of damaged mitochondria. In TG2-ablated cells we observed the presence of a large number of fragmented mitochondria that display decreased membrane potential, downregulation of IF1 along with increased Drp1 and PINK1 levels, two key proteins regulating the mitochondrial fission. Of note, we demonstrate that in healthy mitochondria, TG2 interacts with the dynamic proteins Drp1 and Fis1; interestingly, their interaction is largely reduced upon induction of the fission process by carbonyl cyanide m-chlorophenyl hydrazine (CCCP). In keeping with these findings, mitochondria lacking TG2 are more susceptible to CCCP treatment. As a consequence of accumulation of damaged mitochondria, cells lacking TG2 increased their aerobic glycolysis and became sensitive to the glycolytic inhibitor 2-deoxy-D-glucose (2-DG). In contrast, TG2-proficient cells are more resistant to 2-DG-induced apoptosis as the caspase 3 is inactivated through the enzyme's crosslinking activity. The data presented in this study show that TG2 plays a key role in cellular dynamics and consequently influences the energetic metabolism.

Serotonin and acetylcholine release response in the rat hippocampus during a spatial memory task.

By using in vivo microdialysis we monitored the extracellular levels of acetylcholine and serotonin in the hippocampus of rats performing a spatial memory task. After rats were trained for 10 consecutive days to master a food-reinforced radial-arm maze task, they were implanted with a microdialysis probe in the dorsal hippocampus. On day 12, rats were tested in the maze and acetylcholine and serotonin outputs were monitored before the test, during the waiting phase and while performing the trials. In trained, food-rewarded rats, hippocampal acetylcholine levels increased during the waiting period (181 +/- 90 of baseline) and further increased during the radial-maze performance to 236 +/- 13% of baseline values, while serotonin levels did not change during the waiting period but increased to 142 +/- 3% during the maze performance. To discriminate whether the increase of acetylcholine and serotonin levels during the testing was associated with memory performance or with food consumption, we monitored hippocampal acetylcholine and serotonin release in rats that were trained, but not food rewarded, or in rats that were not trained, but rewarded only on the test day. In the trained, non-rewarded group, acetylcholine release increased during the waiting phase to 168 +/- 6%, but did not increase further during the task performance. In contrast, no change in serotonin release was observed in this group in any phase of the test. In rats which were not trained, but food rewarded, acetylcholine increased only during the maze period (150 +/- 5%). Serotonin increased gradually and become significant at the end of the trials. (130 +/- 3%). While both neurotransmitters could be implicated in feeding behaviour, only activation of cholinergic neurotransmission appears to be associated with memory function. Our results support the following hypotheses: (i) hippocampal acetylcholine could be involved in attentional and cognitive functions underlying motivational processes; (ii) serotonin could be implicated in non-cognitive processes (i.e. in the control of motor and feeding behaviour). Since serotonin and acetylcholine neurotransmission is simultaneously activated during the spatial memory task, this suggests that these neurotransmitter systems regulate behavioural and cognitive functions.

Vitamin A deficiency produces spatial learning and memory impairment in rats.

Vitamin A and its derivatives (retinoids) play important roles in many physiological processes. The recent finding of high levels of cellular retinol-binding protein type 1 immunoreactivity, cellular retinoic acid-binding protein type 1 immunoreactivity and the presence of nuclear retinoid receptors in the central nervous system of adult rodents suggests that retinoids may carry out important roles in the adult brain. In consideration of the role of the hippocampus in spatial learning and memory we evaluated the effect of vitamin A deprivation in adult rats on these functions. Following 12 weeks of vitamin A-free diet, rats were trained to acquire a radial-arm maze task. Results show that this diet induced a severe deficit in the spatial learning and memory task. The cognitive impairment was fully restored when vitamin A was replaced in the diet. We also found a significant decrease in hippocampal acetylcholine release induced by scopolamine, assessed using microdialysis technique, and a reduction in the size of hippocampal nuclei of CA1 region in vitamin-deficient rats, compared to rats fed with a vitamin A-sufficient diet. These results demonstrate that vitamin A has a critical role in the learning and memory processes linked to a proper hippocampal functioning.

Hippocampal acetylcholine release correlates with spatial learning performance in freely moving rats.

To assess the activity of septohippocampal cholinergic neurons during the learning of a radial-arm maze task we measured changes in extracellular acetylcholine levels in the hippocampus by means of the vertical microdialysis technique. During the 12 days spent learning the spatial task the extracellular concentration of acetylcholine in the hippocampus was monitored while rats performed the test. One week before radial-arm maze training a guide cannula was implanted unilaterally in the hippocampus. On each day of testing a removable microdialysis probe was inserted through the guide cannula and the dialysate was collected during the test performance. The concentration of acetylcholine in the dialysate was detected by means of a high-performance liquid chromatograph coupled to an electrochemical detector. We found that hippocampal acetylcholine release progressively increased from 139% to 245% during the 12 days of radial-maze learning and the magnitude of change in acetylcholine output was positively correlated with spatial memory performance, thus suggesting that changes in the functioning of these neurons are involved in learning.

Long-term voluntary ethanol consumption affects neither spatial nor passive avoidance learning, nor hippocampal acetylcholine release in alcohol-preferring rats.

Long-term ethanol consumption in humans and laboratory animals is associated with morphological and functional alterations of brain structures involved in cognitive processes. In the present experiments, we assessed whether voluntary long-term consumption of ethanol by alcohol-preferring (sP) rats under free choice condition with water (also) caused alterations in memory performance and hippocampal acetylcholine (ACh) release in vivo. A group of sP rats were offered a 10% v/v ethanol solution in a free choice with water for 36 weeks; controls had only tap water available. After withdrawal of ethanol, rats were tested in one trial passive avoidance test and thereafter were trained in a food-reinforced radial arm maze task for 12 days. One day after the last session in the radial-arm maze, rats were implanted with a microdialysis probe in the dorsal hippocampus and dialysate concentrations of ACh were measured. No significant differences were observed between sP drinking and control rats in retention latencies in the passive avoidance test, in radial arm-maze performance or in basal levels of hippocampal ACh release. These results show that long-term ethanol consumption by sP rats is not associated with cognitive impairments or with alterations in the hippocampal cholinergic function. To the extent that chronic ethanol intoxication can be considered a causal factor in the development of memory and neurochemical alterations, these results suggest that sP rats self-regulate ethanol consumption so as to avoid intoxication. These findings may challenge the notion that sP rat lines can be considered a valid model of human alcoholism.

A tryptophan-free diet markedly reduces frontocortical 5-HT release, but fails to modify ethanol preference in alcohol-preferring (sP) and non-preferring (sNP) rats.

It has been hypothesised that rat lines genetically selected for their alcohol preference consume large amounts of ethanol because they have a low 5-HT content. Since brain tryptophan (TRP) availability controls the rate at which neurons synthesise and release serotonin (5-HT), we assessed whether the administration of a TRP-supplemented or TRP-free diet for 3 consecutive days influenced alcohol intake in alcohol-preferring and non-preferring sP and sNP rats, respectively. In the same animals extracellular 5-HT concentration was monitored by microdialysis in the frontal cortex. A TRP-free diet progressively and markedly decreased cortical extracellular 5-HT in sP and sNP rats during the treatment period with respect to a balanced diet. However, the TRP-free diet failed to modify alcohol consumption and preference in sP and sNP rats. The TRP-supplemented diet also failed to alter the intake of alcohol in either group of rats. Therefore, these results do not support a specific role of 5-HT transmission in ethanol intake and preference in sP and sNP rats.

The decrease of serotonin release induced by a tryptophan-free amino acid diet does not affect spatial and passive avoidance learning.

We assessed whether consumption of a diet lacking in tryptophan (TRP) resulted in alteration in learning and memory performance and hippocampal 5-HT release in rats. Two hours after the acute administration of TRP-free (T) and balanced (B) diet rats were trained in a one-trial passive avoidance task. The two groups of rats showed no significant difference in retention latencies. Two other groups of rats, fed with the above diets during the acquisition of a radial-arm maze task, showed no difference in baseline performance. The acute ingestion of the T diet produced a significant and long lasting decrease of hippocampal and cortical 5-HT release in rats when compared to the B diet, while the 12th day of the T diet, 5-HT was not detectable in the dialysate. These data indicate that the diminished brain release of 5-HT induced by a T diet is not sufficient to impair cognitive processes.

A physiological method to selectively decrease brain serotonin release.

In this protocol the effect of both an acute and chronic tryptophan (TRP)-free diet on brain serotonin (5-HT) release was studied in rats. Extracellular levels of cortical 5-HT, assessed by in vivo microdialysis, revealed a decrease in the release of this monoamine. Indeed, 120 min after the acute administration of a TRP-free diet, cortical 5-HT release decreased significantly by about 40% with respect to a balanced diet and the decrease persisted for more than 6 h. The chronic intake of a TRP-free diet induced a gradual reduction in 5-HT release. Five days after the diet consumption, our HPLC system detected no 5-HT in cortical dialysate. On the contrary, the acute or chronic administration of the TRP-free diet made no significant change in extracellular noradrenaline content in the frontal cortex, suggesting a specific action of the diet on the serotonergic system. Therefore, the administration of a TRP-free amino acid diet offers a non-pharmacological means for effectively decreasing brain 5-HT release. This diet can be used to study the physiological and behavioral effects of reduced brain 5-HT function.

[Cognitive and performance capacity among adolescents living near a lead and zinc smelter]. / Capacità cognitive e di performance in adolescenti residenti in prossimità di una fonderia di piombo e zinco.

To estimate potential early adverse effects on the Central Nervous System (CNS) due to very low exposure to inorganic lead, the Intelligence Quotient (IQ) of adolescents with present blood lead concentrations (PhB) generally below 10 micrograms/dl was measured. We analyzed blood lead concentration and individual IQ of 32 Sardinian adolescents living in Portoscuso, a town 2 Km far from a lead smelter, and of other 32 controls living in S. Antioco, a town about 15 Km far from the same smelter. The Wechsler Intelligence Scale for Children Revised (WISC-R) was administered. The relation between IQ and blood lead concentration was estimated by linear multivariate models adjusting for several potential confounders, such as the educational and socio-economic level of the parents. The blood lead concentration was in average significantly higher in the Portoscuso group compared to controls. The linear model applied to the total population studied (n 64) showed that the blood lead concentration was inversely and significantly associated with IQ, with an extrapolated decline of 1.29 points in total IQ for each microgram/dl increase of blood concentration. According to the recent scientific literature on this topic, results of our pilot study suggest the need to further lower the definition of an elevated blood lead concentration for children to a threshold significantly below 10 micrograms/dl, value till now considered "safe" for the children's CNS.

Large pseudocounts and L2-norm penalties are necessary for the mean-field inference of Ising and Potts models.

The mean-field (MF) approximation offers a simple, fast way to infer direct interactions between elements in a network of correlated variables, a common, computationally challenging problem with practical applications in fields ranging from physics and biology to the social sciences. However, MF methods achieve their best performance with strong regularization, well beyond Bayesian expectations, an empirical fact that is poorly understood. In this work, we study the influence of pseudocount and L(2)-norm regularization schemes on the quality of inferred Ising or Potts interaction networks from correlation data within the MF approximation. We argue, based on the analysis of small systems, that the optimal value of the regularization strength remains finite even if the sampling noise tends to zero, in order to correct for systematic biases introduced by the MF approximation. Our claim is corroborated by extensive numerical studies of diverse model systems and by the analytical study of the m-component spin model for large but finite m. Additionally, we find that pseudocount regularization is robust against sampling noise and often outperforms L(2)-norm regularization, particularly when the underlying network of interactions is strongly heterogeneous. Much better performances are generally obtained for the Ising model than for the Potts model, for which only couplings incoming onto medium-frequency symbols are reliably inferred.

A role for intracellular zinc in glioma alteration of neuronal chloride equilibrium.

Glioma patients commonly suffer from epileptic seizures. However, the mechanisms of glioma-associated epilepsy are far to be completely understood. Using glioma-neurons co-cultures, we found that tumor cells are able to deeply influence neuronal chloride homeostasis, by depolarizing the reversal potential of γ-aminobutyric acid (GABA)-evoked currents (EGABA). EGABA depolarizing shift is due to zinc-dependent reduction of neuronal KCC2 activity and requires glutamate release from glioma cells. Consistently, intracellular zinc loading rapidly depolarizes EGABA in mouse hippocampal neurons, through the Src/Trk pathway and this effect is promptly reverted upon zinc chelation. This study provides a possible molecular mechanism linking glioma invasion to excitation/inhibition imbalance and epileptic seizures, through the zinc-mediated disruption of neuronal chloride homeostasis.

From rainfall to throughfall in a maritime vineyard.

This study deals with the characteristics of throughfall produced by vine (Vitis vinifera L.) in one of the most common pedoclimatic conditions for grape production: a soil derived from marine sediments under a temperate Mediterranean climate, and located rather close to the seacoast. To distinguish the contribution of the plant from that of the atmospheric deposition, the throughfall was collected for more than one year under real and artificial (plastic) vines; for the same period, also the bulk precipitation was collected. The solution collected were analysed for pH, electrical conductivity, and concentration of cations and anions. For each event, the ionic fluxes of bulk precipitation and throughfall were calculated. Results indicated that the chemical composition of the bulk precipitation was strongly influenced by the proximity of the seashore and, to a lesser extent, by local anthropic activities and windblown material coming from distant areas. The chemical composition of the throughfall was affected by the same factors of bulk precipitation, but also by solubilisation of dry deposition trapped by the canopies, agronomic practices, plant, and living-on-the-leaves microorganisms. The comparison of the characteristics of the throughfall of the real with the artificial vines revealed that the vines are a source of Mg and K. During winter season, the reduction of Ca, NH(4) and PO(4) from bulk precipitation to throughfall was ascribed to the formation of biogenic minerals on the plant surface. The presence of these minerals was proved by X-ray diffraction on the powders collected during the winter season on the surface of cordons and fruiting canes. We conclude that an approach to the estimation of the nutritional potentiality of the soil that includes the contribution of the throughfall is functional to the management of the agro-ecosystem.

High-dimensional inference with the generalized Hopfield model: principal component analysis and corrections.

We consider the problem of inferring the interactions between a set of N binary variables from the knowledge of their frequencies and pairwise correlations. The inference framework is based on the Hopfield model, a special case of the Ising model where the interaction matrix is defined through a set of patterns in the variable space, and is of rank much smaller than N. We show that maximum likelihood inference is deeply related to principal component analysis when the amplitude of the pattern components ξ is negligible compared to √N. Using techniques from statistical mechanics, we calculate the corrections to the patterns to the first order in ξ/√N. We stress the need to generalize the Hopfield model and include both attractive and repulsive patterns in order to correctly infer networks with sparse and strong interactions. We present a simple geometrical criterion to decide how many attractive and repulsive patterns should be considered as a function of the sampling noise. We moreover discuss how many sampled configurations are required for a good inference, as a function of the system size N and of the amplitude ξ. The inference approach is illustrated on synthetic and biological data.