Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Perturbation data”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 109 records · Page 6Linked to original sources

T lymphocytes activated by persistent viral infection differentially modify the expression of metalloproteinases and their endogenous inhibitors, TIMPs, in human astrocytes: relevance to HTLV-I-induced neurological disease.

Activation of T lymphocytes by human pathogens is a key step in the development of immune-mediated neurologic diseases. Because of their ability to invade the CNS and their increased secretion of proinflammatory cytokines, activated CD4+ T cells are thought to play a crucial role in pathogenesis. In the present study, we examined the expression of inflammatory mediators the cytokine-induced metalloproteinases (MMP-2, -3, and -9) and their endogenous inhibitors, tissue inhibitors of metalloproteinases (TIMP-1, -2, and -3), in human astrocytes in response to activated T cells. We used a model system of CD4+ T lymphocytes activated by persistent viral infection (human T lymphotropic virus, HTLV-I) in transient contact with human astrocytes. Interaction with T cells resulted in increased production of MMP-3 and active MMP-9 in astrocytes despite increased expression of endogenous inhibitors, TIMP-1 and TIMP-3. These data suggest perturbation of the MMP/TIMP balance. These changes in MMP and TIMP expression were mediated, in part, by soluble factors (presumably cytokines) secreted by activated T cells. Integrin-mediated cell adhesion is also involved in the change in MMP level, since blockade of integrin subunits (alpha1, alpha3, alpha5, and beta1) on T cells resulted in less astrocytic MMP-9-induced expression. Interestingly, in CNS tissues from neurological HTLV-I-infected patients, MMP-9 was detected in neural cells within the perivascular space, which is infiltrated by mononuclear cells. Altogether, these data emphasize the importance of the MMP-TIMP axis in the complex interaction between the CNS and invading immune cells in the context of virally mediated T cell activation.

Adult↗

[Beta-lactoglobulin AB fluorescence under different physico-chemical conditions. Denaturation by urea and organic solvents].

Dependences of different fluorescence parameters of bovine beta-lactoglobulin AB on the concentrations of urea (pH 2.8-8.8), ethanol (pH 2.1-10.2), and dioxane (pH 5.3) have been investigated. The denaturation properties (the free energy and the stoichiometry of denaturative interaction) are highly dependent on pH values. The data obtained indicate that the hydrophobic interactions are the determining forces in the stabilization process of the beta-lactoglobulin molecule. The relative contribution of these interactions lowers with pH rise. The denaturation of beta-lactoglobulin AB proceeds through two stages under conditions when the protein octamer exists. Up to 30 vol.% of ethanol and dioxane, the penetration of the organic molecules into the external parts of the protein globule takes place. At the concentration of the solvent exceeding 50 vol.% structural transitions are observed. The comparison of fluorescence and perturbation spectral data enables one to localise tryptophan residues in the protein more precisely. The results of this and former reports lead to hypothesis that beta-lactoglobulin may serve as a transporter of some substances which are unstable to acidic media.

Animals↗

Prediction of arterial blood gases by transcutaneous O2 and CO2 in critically ill hyperdynamic trauma patients.

The management of severe adult respiratory distress syndrome in critically injured patients requires the frequent measurement of arterial blood gases for adjustment of cardiovascular and ventilatory support. Since these require blood withdrawal and laboratory determinations, a noninvasive method of assessment of arterial gas tensions would permit more frequent assessment of the patient as well as permitting rapid changes in the patient's ventilatory status to be detected earlier in the clinical course. The role of transcutaneous O2 and CO2 tension in providing these measurements was evaluated in 92 studies in 38 critically ill patients with ARDS due to trauma and/or sepsis. All patients were normodynamic or hyperdynamic at the time of study (cardiac index 2.5 to 7.6 L/min/m2) and were intubated and on increased inspired oxygen fractions (FIO2 = 30 to 100%) delivered by mechanical ventilation, had a range of body temperature from 35.0 to 39.5 degrees C and pH from 7.29 to 7.57 The data from a transcutaneous O2 and CO2 sensor applied to the skin of the anterior thorax were analyzed by multiple regression analysis of variances. Prediction of the arterial oxygen tension (PaO2) from 52 to 253 torr was possible from regression-corrected measurements of the transcutaneous O2 (TcO2): [PaO2 = 1.1 (TcO2) - 0.28 (FIO2) + 45.5]. The arterial carbon dioxide tension (PaCO2) from 26 to 57 torr was predicted from the transcutaneous CO2 (TcCO2):[PaCO2 = 0.76 (TcCO2) + 0.06 (FIO2) + 0.035 (TcO2) + 4.1]. With these corrections, a noninvasive Respiratory Index was computed for assessing ARDS severity, and dynamic changes in arterial gases could be followed in response to postural changes, ventilatory alterations, or cardiovascular perturbations. These data suggest that a reasonable estimate of the arterial blood gases can be obtained from a regression-corrected measurement of the transcutaneous O2 and CO2 tensions in critically injured normodynamic or hyperdynamic ARDS patients.

Adult↗

Tissue resistivity estimation in the presence of positional and geometrical uncertainties.

Geometrical uncertainties (organ boundary variation and electrode position uncertainties) are the biggest sources of error in estimating electrical resistivity of tissues from body surface measurements. In this study, in order to decrease estimation errors, the statistically constrained minimum mean squared error estimation algorithm (MiMSEE) is constrained with a priori knowledge of the geometrical uncertainties in addition to the constraints based on geometry, resistivity range, linearization and instrumentation errors. The MiMSEE calculates an optimum inverse matrix, which maps the surface measurements to the unknown resistivity distribution. The required data are obtained from four-electrode impedance measurements, similar to injected-current electrical impedance tomography (EIT). In this study, the surface measurements are simulated by using a numerical thorax model. The data are perturbed with additive instrumentation noise. Simulated surface measurements are then used to estimate the tissue resistivities by using the proposed algorithm. The results are compared with the results of conventional least squares error estimator (LSEE). Depending on the region, the MiMSEE yields an estimation error between 0.42% and 31.3% compared with 7.12% to 2010% for the LSEE. It is shown that the MiMSEE is quite robust even in the case of geometrical uncertainties.

Algorithms↗

Relaxation spectra of yeast hexokinases. Isomerization of the enzyme.

Yeast hexokinase isozymes P1 and P11 exhibit a pH dependent, rapid relaxation process at 15 degrees C at enzyme concentrations of 100-474 muM and over a pH range of 6-8. The process was detected by equilibrium temperature jump spectroscopy using the indicator probe phenol red. The value of 1/tau varies from about 6 ms-1 at pH 8 for both isozymes to 50 ms-1 for P1 and 85 ms-1 for P11 at pH 6. The data are consistent with a mechanism involving an enzyme isomerization coupled to an ionization. The forward rate constant for the isomerization of the proposed mechanism varies between 3 and 7 ms-1; the ratio of the reverse rate constant to the ionization Ka is between 0.5 and 2 X 10(11) M-1 S-1; the estimated pKa varies between 5.5 and 6.1. The ranges of values in rate constants and pKa represent variations observed between preparations of the same isozyme and between isozymes. The isomerization rate is at least 50 times faster than catalysis under all conditions and the pKa is lower than that controlling activity. The rate of isomerization is unchanged by addition of sugar and nucleotide ligands, but the amplitude of the process is perturbed. These data imply that isomerizing and ionizing forms are sensitive to events at the active site. These equilibria between forms of hexokinase are fast enough, and have the right properties, to be important to the mechanism and regulation of the enzyme.

Adenosine Triphosphate↗

Combined MEG and EEG source imaging by minimization of mutual information.

Though very frequently assumed, the necessity to operate a joint processing of simultaneous magnetoencephalography (MEG) and electroencephalography (EEG) recordings for functional brain imaging has never been clearly demonstrated. However, the very last generation of MEG instruments allows the simultaneous recording of brain magnetic fields and electrical potentials on the scalp. But the general fear regarding the fusion between MEG and EEG data is that the drawbacks from one modality will systematically spoil the performances of the other one without any consequent improvement. This is the case for instance for the estimation of deeper or radial sources with MEG. In this paper, we propose a method for a cooperative processing of MEG and EEG in a distributed source model. First, the evaluation of the respective performances of each modality for the estimation of every dipole in the source pattern is made using a conditional entropy criterion. Then, the algorithm operates a preprocessing of the MEG and EEG gain matrices which minimizes the mutual information between these two transfer functions, by a selective weighting of the MEG and EEG lead fields. This new combined EEG/MEG modality brings major improvements to the localization of active sources, together with reduced sensitivity to perturbations on data.

Bayes Theorem↗

Kinetic analysis of 25-hydroxyvitamin D3 metabolism in strontium-induced rickets in the chick.

Kinetic data analysis was used to derive a six-compartment computer model which describes the in vivo [3H]25-hydroxyvitamin D3 ([3H]25-OHD3) metabolism in control and strontium rachitic chicks. Plasma concentrations of 25-OHD3 (13 pmol/ml) and 25, 25-dihydroxyvitamin D3 (0.9 pmol/ml) were 18 and 125% greater than controls, respectively, whereas the corresponding level for 1alpha,25-dihydroxyvitamin D3 (0.3 pmol/ml) was only 30% of control. Plasma disappearance of 25-HOD3 was fitted using a two-compartment model in which the metabolite extrapolated half-life was nearly twice as large for strontium rachitic chicks (71 compared to 41 h). Intestinal sequestration of 1alpha,25-dihydroxyvitamin D3 was assumed to be irreversible and was fitted by a single exponential term in which metabolite uptake rate and tissue concentration in strontium rickets was suppressed to 20 and 10% of control, respectively. In contrast, uptake of 25-OHD3 by the intestine was observed to occur by a reversible process in which metabolite concentration was 45% greater in the strontium rachitic compared to control group. The developed compartment model accepts time-dependent control or perturbed metabolite data for the plasma and (or) intestinal pools and provides quantitative values for metabolite pool size, flux rate, and turnover time.

Animals↗

Differential contributions of Ng-CAM and N-CAM to cell adhesion in different neural regions.

Individual neurons can express both the neural cell adhesion molecule (N-CAM) and the neuron-glia cell adhesion molecule (Ng-CAM) at their cell surfaces. To determine how the functions of the two molecules may be differentially controlled, we have used specific antibodies to each cell adhesion molecule (CAM) to perturb its function, first in brain membrane vesicle aggregation and then in tissue culture assays testing the fasciculation of neurite outgrowths from cultured dorsal root ganglia, the migration of granule cells in cerebellar explants, and the formation of histological layers in the developing retina. Our strategy was initially to delineate further the binding mechanisms for each CAM. Antibodies to Ng-CAM and N-CAM each inhibited brain membrane vesicle aggregation but the binding mechanisms of the two CAMs differed. As expected from the known homophilic binding mechanism of N-CAM, anti-N-CAM-coated vesicles did not co-aggregate with uncoated vesicles. Anti-Ng-CAM-coated vesicles readily co-aggregated with uncoated vesicles in accord with a postulated heterophilic binding mechanism. It was also shown that N-CAM was not a ligand for Ng-CAM. In contrast to assays with brain membrane vesicles, cellular systems can reveal functional differences for each CAM reflecting its relative amount (prevalence modulation) and location (polarity modulation). Consistent with this, each of the three cellular processes examined in vitro was preferentially inhibited only by anti-N-CAM or by anti-Ng-CAM antibodies. Both neurite fasciculation and the migration of cerebellar granule cells were preferentially inhibited by anti-Ng-CAM antibodies. Anti-N-CAM antibodies inhibited the formation of histological layers in the retina. The data on perturbation by antibodies were correlated with the relative levels of expression of Ng-CAM and N-CAM in each of these different neural regions. Quantitative immunoblotting experiments indicated that the relative Ng-CAM/N-CAM ratios in comparable extracts of brain, dorsal root ganglia, and retina were respectively 0.32, 0.81, and 0.04. During culture of dorsal root ganglia in the presence of nerve growth factor, the Ng-CAM/N-CAM ratio rose to 4.95 in neurite outgrowths and 1.99 in the ganglion proper, reflecting both polarity and prevalence modulation. These results suggest that the relative ability of anti-Ng-CAM and anti-N-CAM antibodies to inhibit cell-cell interactions in different neural tissues is strongly correlated with the local Ng-CAM/N-CAM ratio.(ABSTRACT TRUNCATED AT 400 WORDS)

Animals↗

Localized calcium signaling in multinucleated osteoclasts.

Localized intracellular Ca2+ ([Ca2+]i) pulses, fluctuations, and repetitive spikes were detected in multinucleated rabbit osteoclasts in the presence of serum and in response to calcitonin using the fluorescent calcium indicator fluo-3 and a laser scanning microscope. We observed that these [Ca2+], changes were often restricted within a region of the cell body or propagated from the initial region of occurrence to other parts of the cell body but not to all parts. These observations suggest the existence of significant barriers to Ca2+ transport between different cytoplasmic regions of the osteoclast. To further investigate this phenomenon, we mechanically perturbed different cellular regions by touching locally with a micropipette. This usually induced a local increase in cytosolic and nuclear free [Ca2+]i. In some cases there was propagation of the [Ca2+]i increase to other regions but with part of the cell body not affected. Those regions of the cell body to which the [Ca2+]i increase did not propagate had a [Ca2+]i response to a direct mechanical perturbation. Our data show that osteoclasts can have different [Ca2+]i activities in apparently equivalent cellular regions, no matter how generated. This suggests that there can be a number of spatially separate Ca2+ regulatory systems within an osteoclast cell body.

Animals↗

Isolation and characterization of deteriosomes from rat liver.

Deteriosomes, a new class of microvesicles, have been isolated from rat liver tissue. These microvesicles are similar to those isolated previously from plant tissue [Yao et al., Proc Natl Acad Sci USA 88:2269-2273, 1991] in that they are nonsedimentable and enriched in membrane catabolites, particularly products of phospholipid degradation. Liver deteriosomes range in size from 0.05 microns to 0.11 microns in radius. They are also much more permeable than microsomal membrane vesicles indicating that the deteriosome bilayer is perturbed. The data are consistent with the proposal that deteriosomes are formed from membranes by microvesiculation and that they represent an intermediate stage of membrane deterioration. Furthermore, liver deteriosomes were found to contain phospholipase A2 activity. This suggests that they not only serve as a means of moving destabilizing macromolecular catabolites out of membranes into the cytosol but also possess enzymatic activity. The fact that the specific activity of phospholipase A2 is higher in deteriosomes than in deteriosome-free cytosol suggests that some of the enzymatic activity traditionally assumed to be cytosolic may in fact be associated with deteriosomes.

Animals↗

Adaptation of reach-to-grasp movement in response to force perturbations.

This study examined how reach-to-grasp movements are modified during adaptation to external force perturbations applied on the arm during reach. Specifically, we examined whether the organization of these movements was dependent upon the condition under which the perturbation was applied. In response to an auditory signal, all subjects were asked to reach for a vertical dowel, grasp it between the index finger and thumb, and lift it a short distance off the table. The subjects were instructed to do the task as fast as possible. The perturbation was an elastic load acting on the wrist at an angle of 105 deg lateral to the reaching direction. The condition was modified by changing the predictability with which the perturbation was applied in a given trial. After recording unperturbed control trials, perturbations were applied first on successive trials (predictable perturbations) and then were applied randomly (unpredictable perturbations). In the early predictable perturbation trials, reach path length became longer and reaching duration increased. As more predictable perturbations were applied, the reach path length gradually decreased and became similar to that of control trials. Reaching duration also decreased gradually as the subjects adapted by exerting force against the perturbation. In addition, the amplitude of peak grip aperture during arm transport initially increased in response to repeated perturbations. During the course of learning, it reached its maximum and thereafter slightly decreased. However, it did not return to the normal level. The subjects also adapted to the unpredictable perturbations through changes in both arm transport and grasping components, indicating that they can compensate even when the occurrence of the perturbation cannot be predicted during the inter-trial interval. Throughout random perturbation trials, large grip aperture values were observed, suggesting that a conservative aperture level is set regardless of whether the reaching arm is perturbed or not. In addition, the results of the predictable perturbations showed that the time from movement onset to the onset of grip aperture closure changed as adaptation occurred. However, the spatial location where the onset of finger closure occurred showed minimum changes with perturbation. These data suggest that the onset of finger closure is dependent upon distance to target rather than the temporal relationship of the grasp relative to the transport phase of the movement.

Adaptation, Physiological↗

Control model of human stance using fuzzy logic.

A control model of human stance is proposed based on knowledge from behavioral experiments and physiological systems. The proposed model is based on the control of global variables specific to body orientation and alignment, rather than on the control of the body's center of mass within the base of support. Furthermore, the proposed control model is not based on purely inverted pendulum body mechanics where only motion at one joint is controlled, as for instance the ankle. In the proposed model, the degrees of freedom are controlled by using reciprocal and synergistic muscle actions at multiple joints. The control model is based on three sets of different global variables which act in parallel: (1) limb length and its derivative, (2) limb orientation and its derivative, and (3) trunk attitude and its derivative. An important feature of the control model is the use of fuzzy logic, which enables us to model experimental findings and physiological knowledge in a meaningful and explicit way using fuzzy if-then rules. In the control model, 36 fuzzy if-then rules are implemented and applied using a four-linked segment model consisting of a trunk, thigh, shank and foot. Uni- and biarticular limb muscles and trunk muscles are represented as torque actuators at each individual joint. In the model, three sets of global variables act in parallel and make corrective and coordinated responses to internal, self-induced perturbations. The data show that the use of global variables and fuzzy logic successfully enables us to model human standing with sway about a point of equilibrium. Small changes in, for example, total body sway are comparable to those seen during natural sway in human stance. The selected controllers--limb length, limb orientation and trunk attitude--seem to be appropriate for human stance control.

Fuzzy Logic↗

Glucose utilization and insulin binding in discrete brain areas of obese rats.

The present study was carried out to determine whether genetically obese Zucker rats present changes in brain glucose utilization and/or insulin binding when compared to their lean counterparts. Glucose utilization in the whole brain, determined by measurement of 2-deoxy(1-3H)glucose-6-phosphate, was significantly lower in obese than in lean Zucker rats. In order to precise the structure involved, we then used quantitative autoradiography methods after either (1-14C) 2-deoxyglucose injection or 125I-insulin incubation. In obese rats, local cerebral glucose utilization (LCGU) was significantly decreased in the external plexiform layer (-37%, p < 0.05), in the lateral hypothalamus (-23%, p < 0.05), and in the basolateral amygdaloid nucleus (-30%, p < 0.05). In contrast, no difference in specific insulin binding was found between the two genotypes in any of the areas studied. These results are consistent with some data showing a decrease of LCGU in hyperinsulinemic rats. All together, these data show perturbations of glucose utilization, particularly in structures linked to the regulation of body weight and food intake in obese Zucker rats.

Amygdala↗

Interaction of methylmercury chloride with cellular energetics and related processes.

Upon exposure to methylmercury chloride, the whole-cell oxygen uptake by the yeast Saccharomyces cerevisiae ceases. On a fermentable carbon source, carbon dioxide continues to be evolved after respiration has stopped, indicating that fermentation is still active. Dextrose and glycerol uptake also persists until the respective processes, fermentation and respiration, are totally inhibited. Protein and nucleic acid synthesis are blocked with similar concentrations of methylmercury, while cytochrome c, the terminal component of the electron transport chain, is unaltered by the toxicant. Surprisingly, the intracellular ATP is higher in the treated cells than in the controls, although they eventually fall in response to higher concentrations of methylmercury, while cytochrome c, the terminal component of the electron transport chain, is unaltered by the toxicant. Surprisingly, the intracellular ATP is higher in the treated cells than in the controls, although they eventually fall in response to higher concentration or longer exposure. High-pressure liquid chromatography profiles show that the amounts of the other nucleotides are either unaltered or increased. The entire inhibitory process is reversible with time or fresh medium at low methylmercury concentrations. These results do not support the hypothesis expressed by several authors of an inhibition of ATP biosynthesis resulting from membrane perturbation. These data suggest that the decrease in ATP--when induced by the organomercurial--is a secondary process and is not the result of direct mitochondrial toxicity.

Adenosine Triphosphate↗

Hemopoietic stem cell heterogeneity: use of cell cycle-specific drugs to look for age-associated alterations.

Hemopoietic tissue is vulnerable to perturbations, and data show that it is an appropriate tissue in which to look for age-associated alterations. This tissue has a high regenerative capacity, is composed of a heterogeneous population of stem cells that are capable of self renewal or differentiation, or both, and is sustained by a pool of resting cells. The heterogeneity of bone marrow has made characterization of the cellular elements difficult. Techniques commonly used to identify and quantify the various maturation levels of hemopoietic stem cells and the limitations of these techniques are discussed. Most techniques used to assay age-associated changes in bone marrow have not differentiated between specific cellular alterations or shifts in the distribution of the cellular elements. In particular, it has been difficult to determine the stability of the non-dividing stem cell because of the low incidence of this cell (6 per 1000) and the lack of a specific assay for this important cell type. The use of cell cycle-specific drugs has provided quantitative information on specific subpopulations of hemopoietic stem cells and seems to be the most promising approach towards determining qualitative and quantitative differences in the hemopoietic stem cells of young and old individuals.

Aging↗

Conformational changes induced by hapten in murine monoclonal antibodies to dinitrophenyl groups--the analysis by temperature-perturbation spectroscopy.

Two samples of murine monoclonal antibodies to dinitrophenyl groups were studied by difference thermal perturbation spectroscopy with particular attention to changes in the amount of perturbed chromophores induced in antibodies as a result of hapten binding (epsilon-2,4-dinitrophenyl-L-lysine). Despite the fact that both antibody samples belong to immunoglobulin G1 and have the same type of light chain, kappa, they were found to differ significantly in the number of the chromophores perturbed by temperature. The binding of hapten decreases the perturbation of chromophores only in the sample with the less rigid structure, as regards thermal perturbation. These data provide evidence that differences in the rigidity of the structure of variable domains affect the extent of conformational changes induced in the antibodies due to the interaction with an antigen.

Animals↗

Studies on the enzymatic and physicochemical behaviour of the trichloroacetic acid-treated and untreated bovine pancreatic ribonuclease.

Exposure of ribonuclease A to 5% trichloroacetic acid inactivates the enzyme partially. One of the possible reasons for such inactivation might be the exposure of one of the buried tyrosyl groups to the outside surface of the molecule (Sagar and Pandit (1983) Biochim. Biophys. Acta 743, 303-309). The trichloroacetic acid-treated enzyme hydrolysed 2':3'-cCMP with an efficiency of about 60%; while with rRNA as substrate, it is about 45%. Results indicate that apart from the reduction in the activity on trichloroacetic acid treatment, the enzyme possesses a reduced ability to break down the secondary structures of substrates such as rRNA in the first phase of the reaction. Thermal unfolding of ribonuclease A was followed by various physicochemical techniques such as UV absorbance, CD-spectroscopy and differential scanning microcalorimetry. The results indicate that the enzyme, after trichloroacetic acid-treatment, has a less ordered structure when compared to that of untreated enzyme. Thermal unfolding profiles reveal that trichloroacetic acid-treated ribonuclease A, like the untreated enzyme, follows a one-step transition with relatively lower transition temperature (Tm). NMR-spectral data suggests perturbations in the histidyl environment at the active site.

Binding Sites↗

Disagreement-informed arbitration for gene regulatory network inference: A score-level meta-classifier and a diagnostic typology of inter-method conflict.

Gene regulatory network inference methods routinely disagree about individual edges, and practitioners resolve those conflicts by choosing one method or averaging them all. We ask whether the conflict can instead be arbitrated per edge. A gradient-boosted classifier is trained on the raw scores that ten inference methods-correlation-based, information-theoretic, sparse-regression and tree-ensemble, including GENIE3, GRNBoost2, CLR and ARACNe-assign to each candidate regulator-target pair, so that the weight given to each method varies from edge to edge. Across six single-cell perturbation screens spanning four cell types, arbitration improves on mean ensembling by +0.056 AUROC on Adamson and +0.083 on Shifrut under target-grouped cross-validation. The evaluation protocol turns out to matter more than the model. Edge-level cross-validation, standard in this literature, inflates apparent gains by 0.060 AUROC through target-gene leakage-comparable to the entire honest improvement. The effect is far larger for methods that represent genes implicitly: a supervised graph-attention link predictor trained on identical folds scores AUROC 0.930 under edge-level cross-validation, better than anything else we evaluate, and 0.533 once target genes are held out. Any method that parameterises genes is exposed, which covers most graph- and embedding-based approaches. A five-category typology of inter-method conflict localises where arbitration pays off, with the largest gains on edges where the methods disagree and the smallest where they already agree, while adding nothing as model input; we therefore report it as a diagnostic instrument rather than a modelling contribution. We also characterise what the ground truth measures: most perturbed genes in widely used screens are not transcription factors, and a mediation screen bounds how much of the perturbation response can be direct.

Ensemble methods↗