A MULTILEVEL MODEL TO ADDRESS BATCH EFFECTS IN COPY NUMBER ESTIMATION USING SNP ARRAYS

Submicroscopic changes in chromosomal DNA copy number dosage are common and have been implicated in many heritable diseases and cancers. Recent high-throughput technologies have a resolution that permits the detection of segmental changes in DNA copy number that span thousands of basepairs across the genome. Genome-wide association studies (GWAS) may simultaneously screen for copy number-phenotype and SNP-phenotype associations as part of the analytic strategy. However, genome-wide array analyses are particularly susceptible to batch effects as the logistics of preparing DNA and processing thousands of arrays often involves multiple laboratories and technicians, or changes over calendar time to the reagents and laboratory equipment. Failure to adjust for batch effects can lead to incorrect inference and requires inefficient post-hoc quality control procedures that exclude regions that are associated with batch. Our work extends previous model-based approaches for copy number estimation by explicitly modeling batch effects and using shrinkage to improve locus-specific estimates of copy number uncertainty. Key features of this approach include the use of diallelic genotype calls from experimental data to estimate batch- and locus-specific parameters of background and signal without the requirement of training data. We illustrate these ideas using a study of bipolar disease and a study of chromosome 21 trisomy. The former has batch effects that dominate much of the observed variation in quantile-normalized intensities, while the latter illustrates the robustness of our approach to datasets where as many as 25% of the samples have altered copy number. Locus-specific estimates of copy number can be plotted on the copy-number scale to investigate mosaicism and guide the choice of appropriate downstream approaches for smoothing the copy number as a function of physical position. The software is open source and implemented in the R package CRLMM available at Bioconductor (http:www.bioconductor.org).

A BAYESIAN HIERARCHICAL MODEL FOR CONSTRAINED DISTRIBUTED LAG FUNCTIONS: ESTIMATING THE TIME COURSE OF HOSPITALIZATION ASSOCIATED WITH AIR POLLUTION EXPOSURE

Numerous time series studies have provided strong evidence of an association between increased levels of ambient air pollution and increased levels of hospital admissions, typically at 0, 1, or 2 days after an air pollution episode. An important research aim is to extend existing statistical models so that a more detailed understanding of the time course of hospitalization after exposure to air pollution can be obtained. Information about this time course, combined with prior knowledge about biological mechanisms, could provide the basis for hypotheses concerning the mechanism by which air pollution causes disease. Previous studies have identified two important methodological questions: (1) How can we estimate the shape of the distributed lag between increased air pollution exposure and increased mortality or morbidity? and (2) How should we estimate the cumulative population health risk from short-term exposure to air pollution? Distributed lag models are appropriate tools for estimating air pollution health effects that may be spread over several days. However, estimation for distributed lag models in air pollution and health applications is hampered by the substantial noise in the data and the inherently weak signal that is the target of investigation. We introduce an hierarchical Bayesian distributed lag model that incorporates prior information about the time course of pollution effects and combines information across multiple locations. The model has a connection to penalized spline smoothing using a special type of penalty matrix. We apply the model to estimating the distributed lag between exposure to particulate matter air pollution and hospitalization for cardiovascular and respiratory disease using data from a large United States air pollution and hospitalization database of Medicare enrollees in 94 counties covering the years 1999-2002.

Orthotopic transplantation of v-src-expressing glioma cell lines into immunocompetent mice: establishment of a new transplantable in vivo model for malignant glioma

OBJECT: The aim of this study was to develop and characterize a new orthotopic, syngeneic, transplantable mouse brain tumor model by using the cell lines Tu-9648 and Tu-2449, which were previously isolated from tumors that arose spontaneously in glial fibrillary acidic protein (GFAP)-v-src transgenic mice. METHODS: Striatal implantation of a 1-microl suspension of 5000 to 10,000 cells from either clone into syngeneic B6C3F1 mice resulted in tumors that were histologically identified as malignant gliomas. Prior subcutaneous inoculations with irradiated autologous cells inhibited the otherwise robust development of a microscopically infiltrating malignant glioma. Untreated mice with implanted tumor cells were killed 12 days later, when the resultant gliomas were several millimeters in diameter. Immunohistochemically, the gliomas displayed both the astroglial marker GFAP and the oncogenic form of signal transducer and activator of transcription-3 (Stat3). This form is called tyrosine-705 phosphorylated Stat3, and is found in many malignant entities, including human gliomas. Phosphorylated Stat3 was particularly prominent, not only in the nucleus but also in the plasma membrane of peripherally infiltrating glioma cells, reflecting persistent overactivation of the Janus kinase/Stat3 signal transduction pathway. The Tu-2449 cells exhibited three non-random structural chromosomal aberrations, including a deletion of the long arm of chromosome 2 and an apparently balanced translocation between chromosomes 1 and 3. The GFAP-v-src transgene was mapped to the pericentromeric region of chromosome 18. CONCLUSIONS: The high rate of engraftment, the similarity to the high-grade malignant glioma of origin, and the rapid, locally invasive growth of these tumors should make this murine model useful in testing novel therapies for human malignant gliomas.

Proliferation signal inhibitor-induced decrease of vascular endothelial cadherin expression and increase of endothelial permeability in vitro are prevented by an anti-oxidant

BACKGROUND: Rapamycines, sirolimus (SRL) and everolimus (ERL), are proliferation signal inhibitors (PSIs). PSI therapy often leads to edema. We hypothesized that increased oxidative stress in response to PSIs may modulate the expression of vascular endothelial (VE)-cadherin on endothelial cells (ECs) and, subsequently, vascular permeability, which in turn may be involved in the development of edema. METHODS: Experiments were performed on human umbilical vein ECs (HUVECs). Oxidative stress was measured by dichlorofluorescein-diacetate. Expression of VE-cadherin was evaluated by immunofluorescent staining and western blot analysis. Endothelial "permeability" was assessed using a transwell model. RESULTS: SRL and ERL, at concentrations of 1, 10 and 100 nmol/liter, enhanced oxidative stress (SRL: 24 +/- 12%, 29 +/- 9%, 41 +/- 13% [p < 0.05, in all three cases]; ERL: 13 +/- 10%, 27 +/- 2%, 40 +/- 12% [p < 0.05, in the latter two cases], respectively) on HUVECs, which was inhibited by the anti-oxidant, N-acetyl-cysteine (NAC) and, to a lesser extent, by the specific inhibitor of nitric oxide synthase, N-Omega-nitro-L-arginine methylester. By the use of NAC, VE-cadherin expression remained comparable with control, according to both immunocytochemistry and western blot analysis. Permeability was significantly increased by SRL and ERL at 100 nmol/liter (29.5 +/- 6.4% and 33.8 +/- 4.2%, respectively); however, co-treatment with NAC abrogated the increased permeability. CONCLUSIONS: EC homeostasis, as indicated by VE-cadherin expression, may be damaged by SRL and ERL, but resolved by the anti-oxidant NAC.

Stochastic knock detection model for spark ignited engines

This report presents the development of a Stochastic Knock Detection (SKD) method for combustion knock detection in a spark-ignition engine using a model based design approach. Knock Signal Simulator (KSS) was developed as the plant model for the engine. The KSS as the plant model for the engine generates cycle-to-cycle accelerometer knock intensities following a stochastic approach with intensities that are generated using a Monte Carlo method from a lognormal distribution whose parameters have been predetermined from engine tests and dependent upon spark-timing, engine speed and load. The lognormal distribution has been shown to be a good approximation to the distribution of measured knock intensities over a range of engine conditions and spark-timings for multiple engines in previous studies. The SKD method is implemented in Knock Detection Module (KDM) which processes the knock intensities generated by KSS with a stochastic distribution estimation algorithm and outputs estimates of high and low knock intensity levels which characterize knock and reference level respectively. These estimates are then used to determine a knock factor which provides quantitative measure of knock level and can be used as a feedback signal to control engine knock. The knock factor is analyzed and compared with a traditional knock detection method to detect engine knock under various engine operating conditions. To verify the effectiveness of the SKD method, a knock controller was also developed and tested in a model-in-loop (MIL) system. The objective of the knock controller is to allow the engine to operate as close as possible to its border-line spark-timing without significant engine knock. The controller parameters were tuned to minimize the cycle-to-cycle variation in spark timing and the settling time of the controller in responding to step increase in spark advance resulting in the onset of engine knock. The simulation results showed that the combined system can be used adequately to model engine knock and evaluated knock control strategies for a wide range of engine operating conditions.

Direct MR arthrography at 1.5 and 3.0 T: signal dependence on gadolinium and iodine concentrations--phantom study

PURPOSE: To prospectively quantify in vitro the influence of gadopentetate dimeglumine and ioversol on the magnetic resonance (MR) imaging signal observed with a variety of musculoskeletal pulse sequences to predict optimum gadolinium concentrations for direct MR arthrography at 1.5 and 3.0 T. MATERIALS AND METHODS: In an in vitro study, T1 and T2 relaxation times of three dilution series of gadopentetate dimeglumine (concentration, 0-20.0 mmol gadolinium per liter) at ioversol concentrations with iodine concentration of 0, 236.4, and 1182 mmol iodine per liter (corresponding to 0, 30, and 150 mg of iodine per milliliter) were measured at 1.5 and 3.0 T. The relaxation rate dependence on concentrations of gadolinium and iodine was analytically modeled, and continuous profiles of signal versus gadolinium concentration were calculated for 10 pulse sequences used in current musculoskeletal imaging. After fitting to experimental discrete profiles, maximum signal-to-noise ratio (SNR), gadolinium concentration with maximum SNR, and range of gadolinium concentration with 90% of maximum SNR were derived. The overall influence of field strength and iodine concentration on these parameters was assessed by using t tests. The deviation of simulated from experimental signal-response profiles was assessed with the autocorrelation of the residuals. RESULTS: The model reproduced relaxation rates of 0.37-38.24 sec(-1), with a mean error of 4.5%. Calculated SNR profiles matched the discrete experimental profiles, with autocorrelation of the residuals divided by the mean of less than 5.0. Admixture of ioversol consistently reduced T1 and T2, narrowed optimum gadolinium concentration ranges (P = .004-.006), and reduced maximum SNR (P < .001 to not significant). Optimum gadolinium concentration was 0.7-3.4 mmol/L at both field strengths. At 3.0 T, maximum SNR was up to 75% higher than at 1.5 T. CONCLUSION: Admixture of ioversol to gadopentetate dimeglumine solutions results in a consistent additional relaxation enhancement, which can be analytically modeled to allow a near-quantitative a priori optimized match of contrast media concentrations and imaging protocol for a broad variety of pulse sequences.

Visualization of alveolar recruitment in a porcine model of unilateral lung lavage using 3He-MRI

BACKGROUND: In the acute respiratory distress syndrome potentially recruitable lung volume is currently discussed. (3)He-magnetic resonance imaging ((3)He-MRI) offers the possibility to visualize alveolar recruitment directly. METHODS: With the approval of the state animal care committee, unilateral lung damage was induced in seven anesthetized pigs by saline lavage of the right lungs. The left lung served as an intraindividual control (healthy lung). Unilateral lung damage was confirmed by conventional proton MRI and spiral-CT scanning. The total aerated lung volume was determined both at a positive end-expiratory pressure (PEEP) of 0 and 10 mbar from three-dimensionally reconstructed (3)He images, both for healthy and damaged lungs. The fractional increase of aerated volume in damaged and healthy lungs, followed by a PEEP increase from 0 to 10 mbar, was compared. RESULTS: Aerated gas space was visualized with a high spatial resolution in the three-dimensionally reconstructed (3)He-MR images, and aeration defects in the lavaged lung matched the regional distribution of atelectasis in proton MRI. After recruitment and PEEP increase, the aerated volume increased significantly both in healthy lungs from 415 ml [270-445] (median [min-max]) to 481 ml [347-523] and in lavaged lungs from 264 ml [71-424] to 424 ml [129-520]. The fractional increase in lavaged lungs was significantly larger than that in healthy lungs (healthy: 17% [11-38] vs. lavage: 42% [14-90] (P=0.031). CONCLUSION: The (3)He-MRI signal might offer an experimental approach to discriminate atelectatic vs. poor aerated lung areas in a lung damage animal model. Our results confirm the presence of potential recruitable lung volume by either alveolar collapse or alveolar flooding, in accordance with previous reports by computed tomography.

Simulating the temperature and precipitation signal in an Alpine ice core

Accumulation and delta O-18 data from Alpine ice cores provide information on past temperature and precipitation. However, their correlation with seasonal or annual mean temperature and precipitation at nearby sites is often low. This is partly due to the irregular sampling of the atmosphere by the ice core (i.e. ice cores almost only record precipitation events and not dry periods) and the possible incongruity between annual layers and calendar years. Using daily meteorological data from a nearby station and reanalyses, we replicate the ice core from the Grenzgletscher (Switzerland, 4200m a.s.l.) on a sample-by-sample basis by calculating precipitation-weighted temperature (PWT) over short intervals. Over the last 15 yr of the ice core record, accumulation and delta O-18 variations can be well reproduced on a sub-seasonal scale. This allows a wiggle-matching approach for defining quasi-annual layers, resulting in high correlations between measured quasi-annual delta O-18 and PWT. Further back in time, the agreement deteriorates. Nevertheless, we find significant correlations over the entire length of the record (1938-1993) of ice core delta O-18 with PWT, but not with annual mean temperature. This is due to the low correlations between PWT and annual mean temperature, a characteristic which in ERA-Interim reanalysis is also found for many other continental mid-to-high-latitude regions. The fact that meteorologically very different years can lead to similar combinations of PWT and accumulation poses limitations to the use of delta O-18 from Alpine ice cores for temperature reconstructions. Rather than for reconstructing annual mean temperature, delta O-18 from Alpine ice cores should be used to reconstruct PWT over quasi-annual periods. This variable is reproducible in reanalysis or climate model data and could thus be assimilated into conventional climate models.

Impact of volcanic stratospheric aerosols on diurnal temperature range in Europe over the past 200 years: Observations versus model simulations

We analyze the impact of stratospheric volcanic aerosols on the diurnal temperature range (DTR) over Europe using long-term subdaily station records. We compare the results with a 28-member ensemble of European Centre/Hamburg version 5.4 (ECHAM5.4) general circulation model simulations. Eight stratospheric volcanic eruptions during the instrumental period are investigated. Seasonal all- and clear-sky DTR anomalies are compared with contemporary (approximately 20 year) reference periods. Clear sky is used to eliminate cloud effects and better estimate the signal from the direct radiative forcing of the volcanic aerosols. We do not find a consistent effect of stratospheric aerosols on all-sky DTR. For clear skies, we find average DTR anomalies of −0.08°C (−0.13°C) in the observations (in the model), with the largest effect in the second winter after the eruption. Although the clear-sky DTR anomalies from different stations, volcanic eruptions, and seasons show heterogeneous signals in terms of order of magnitude and sign, the significantly negative DTR anomalies (e.g., after the Tambora eruption) are qualitatively consistent with other studies. Referencing with clear-sky DTR anomalies to the radiative forcing from stratospheric volcanic eruptions, we find the resulting sensitivity to be of the same order of magnitude as previously published estimates for tropospheric aerosols during the so-called “global dimming” period (i.e., 1950s to 1980s). Analyzing cloud cover changes after volcanic eruptions reveals an increase in clear-sky days in both data sets. Quantifying the impact of stratospheric volcanic eruptions on clear-sky DTR over Europe provides valuable information for the study of the radiative effect of stratospheric aerosols and for geo-engineering purposes.

Differential regulation of human 3β-hydroxysteroid dehydrogenase type 2 for steroid hormone biosynthesis by starvation and cyclic AMP stimulation: studies in the human adrenal NCI-H295R cell model

Human steroid biosynthesis depends on a specifically regulated cascade of enzymes including 3β-hydroxysteroid dehydrogenases (HSD3Bs). Type 2 HSD3B catalyzes the conversion of pregnenolone, 17α-hydroxypregnenolone and dehydroepiandrosterone to progesterone, 17α-hydroxyprogesterone and androstenedione in the human adrenal cortex and the gonads but the exact regulation of this enzyme is unknown. Therefore, specific downregulation of HSD3B2 at adrenarche around age 6-8 years and characteristic upregulation of HSD3B2 in the ovaries of women suffering from the polycystic ovary syndrome remain unexplained prompting us to study the regulation of HSD3B2 in adrenal NCI-H295R cells. Our studies confirm that the HSD3B2 promoter is regulated by transcription factors GATA, Nur77 and SF1/LRH1 in concert and that the NBRE/Nur77 site is crucial for hormonal stimulation with cAMP. In fact, these three transcription factors together were able to transactivate the HSD3B2 promoter in placental JEG3 cells which normally do not express HSD3B2. By contrast, epigenetic mechanisms such as methylation and acetylation seem not involved in controlling HSD3B2 expression. Cyclic AMP was found to exert differential effects on HSD3B2 when comparing short (acute) versus long-term (chronic) stimulation. Short cAMP stimulation inhibited HSD3B2 activity directly possibly due to regulation at co-factor or substrate level or posttranslational modification of the protein. Long cAMP stimulation attenuated HSD3B2 inhibition and increased HSD3B2 expression through transcriptional regulation. Although PKA and MAPK pathways are obvious candidates for possibly transmitting the cAMP signal to HSD3B2, our studies using PKA and MEK1/2 inhibitors revealed no such downstream signaling of cAMP. However, both signaling pathways were clearly regulating HSD3B2 expression.

Dynamics of a minimal model of interlocked positive and negative feedback loops of transcriptional regulation by cAMP-response element binding proteins.

cAMP-response element binding (CREB) proteins are involved in transcriptional regulation in a number of cellular processes (e.g., neural plasticity and circadian rhythms). The CREB family contains activators and repressors that may interact through positive and negative feedback loops. These loops can be generated by auto- and cross-regulation of expression of CREB proteins, via CRE elements in or near their genes. Experiments suggest that such feedback loops may operate in several systems (e.g., Aplysia and rat). To understand the functional implications of such feedback loops, which are interlocked via cross-regulation of transcription, a minimal model with a positive and negative loop was developed and investigated using bifurcation analysis. Bifurcation analysis revealed diverse nonlinear dynamics (e.g., bistability and oscillations). The stability of steady states or oscillations could be changed by time delays in the synthesis of the activator (CREB1) or the repressor (CREB2). Investigation of stochastic fluctuations due to small numbers of molecules of CREB1 and CREB2 revealed a bimodal distribution of CREB molecules in the bistability region. The robustness of the stable HIGH and LOW states of CREB expression to stochastic noise differs, and a critical number of molecules was required to sustain the HIGH state for days or longer. Increasing positive feedback or decreasing negative feedback also increased the lifetime of the HIGH state, and persistence of this state may correlate with long-term memory formation. A critical number of molecules was also required to sustain robust oscillations of CREB expression. If a steady state was near a deterministic Hopf bifurcation point, stochastic resonance could induce oscillations. This comparative analysis of deterministic and stochastic dynamics not only provides insights into the possible dynamics of CREB regulatory motifs, but also demonstrates a framework for understanding other regulatory processes with similar network architecture.

Last interglacial temperature evolution — a model inter-comparison

There is a growing number of proxy-based reconstructions detailing the climatic changes that occurred during the last interglacial period (LIG). This period is of special interest, because large parts of the globe were characterized by a warmer-than-present-day climate, making this period an interesting test bed for climate models in light of projected global warming. However, mainly because synchronizing the different palaeoclimatic records is difficult, there is no consensus on a global picture of LIG temperature changes. Here we present the first model inter-comparison of transient simulations covering the LIG period. By comparing the different simulations, we aim at investigating the common signal in the LIG temperature evolution, investigating the main driving forces behind it and at listing the climate feedbacks which cause the most apparent inter-model differences. The model inter-comparison shows a robust Northern Hemisphere July temperature evolution characterized by a maximum between 130–125 ka BP with temperatures 0.3 to 5.3 K above present day. A Southern Hemisphere July temperature maximum, −1.3 to 2.5 K at around 128 ka BP, is only found when changes in the greenhouse gas concentrations are included. The robustness of simulated January temperatures is large in the Southern Hemisphere and the mid-latitudes of the Northern Hemisphere. For these regions maximum January temperature anomalies of respectively −1 to 1.2 K and −0.8 to 2.1 K are simulated for the period after 121 ka BP. In both hemispheres these temperature maxima are in line with the maximum in local summer insolation. In a number of specific regions, a common temperature evolution is not found amongst the models. We show that this is related to feedbacks within the climate system which largely determine the simulated LIG temperature evolution in these regions. Firstly, in the Arctic region, changes in the summer sea-ice cover control the evolution of LIG winter temperatures. Secondly, for the Atlantic region, the Southern Ocean and the North Pacific, possible changes in the characteristics of the Atlantic meridional overturning circulation are crucial. Thirdly, the presence of remnant continental ice from the preceding glacial has shown to be important when determining the timing of maximum LIG warmth in the Northern Hemisphere. Finally, the results reveal that changes in the monsoon regime exert a strong control on the evolution of LIG temperatures over parts of Africa and India. By listing these inter-model differences, we provide a starting point for future proxy-data studies and the sensitivity experiments needed to constrain the climate simulations and to further enhance our understanding of the temperature evolution of the LIG period.

Application of Signal Advance Technology to Electrophysiology

Medical instrumentation used in diagnosis and treatment relies on the accurate detection and processing of various physiological events and signals. While signal detection technology has improved greatly in recent years, there remain inherent delays in signal detection/ processing. These delays may have significant negative clinical consequences during various pathophysiological events. Reducing or eliminating such delays would increase the ability to provide successful early intervention in certain disorders thereby increasing the efficacy of treatment. In recent years, a physical phenomenon referred to as Negative Group Delay (NGD), demonstrated in simple electronic circuits, has been shown to temporally advance the detection of analog waveforms. Specifically, the output is temporally advanced relative to the input, as the time delay through the circuit is negative. The circuit output precedes the complete detection of the input signal. This process is referred to as signal advance (SA) detection. An SA circuit model incorporating NGD was designed, developed and tested. It imparts a constant temporal signal advance over a pre-specified spectral range in which the output is almost identical to the input signal (i.e., it has minimal distortion). Certain human patho-electrophysiological events are good candidates for the application of temporally-advanced waveform detection. SA technology has potential in early arrhythmia and epileptic seizure detection and intervention. Demonstrating reliable and consistent temporally advanced detection of electrophysiological waveforms may enable intervention with a pathological event (much) earlier than previously possible. SA detection could also be used to improve the performance of neural computer interfaces, neurotherapy applications, radiation therapy and imaging. In this study, the performance of a single-stage SA circuit model on a variety of constructed input signals, and human ECGs is investigated. The data obtained is used to quantify and characterize the temporal advances and circuit gain, as well as distortions in the output waveforms relative to their inputs. This project combines elements of physics, engineering, signal processing, statistics and electrophysiology. Its success has important consequences for the development of novel interventional methodologies in cardiology and neurophysiology as well as significant potential in a broader range of both biomedical and non-biomedical areas of application.

Dynamics of a minimal model of interlocked positive and negative feedback loops of transcriptional regulation by cAMP-response element binding proteins.

cAMP-response element binding (CREB) proteins are involved in transcriptional regulation in a number of cellular processes (e.g., neural plasticity and circadian rhythms). The CREB family contains activators and repressors that may interact through positive and negative feedback loops. These loops can be generated by auto- and cross-regulation of expression of CREB proteins, via CRE elements in or near their genes. Experiments suggest that such feedback loops may operate in several systems (e.g., Aplysia and rat). To understand the functional implications of such feedback loops, which are interlocked via cross-regulation of transcription, a minimal model with a positive and negative loop was developed and investigated using bifurcation analysis. Bifurcation analysis revealed diverse nonlinear dynamics (e.g., bistability and oscillations). The stability of steady states or oscillations could be changed by time delays in the synthesis of the activator (CREB1) or the repressor (CREB2). Investigation of stochastic fluctuations due to small numbers of molecules of CREB1 and CREB2 revealed a bimodal distribution of CREB molecules in the bistability region. The robustness of the stable HIGH and LOW states of CREB expression to stochastic noise differs, and a critical number of molecules was required to sustain the HIGH state for days or longer. Increasing positive feedback or decreasing negative feedback also increased the lifetime of the HIGH state, and persistence of this state may correlate with long-term memory formation. A critical number of molecules was also required to sustain robust oscillations of CREB expression. If a steady state was near a deterministic Hopf bifurcation point, stochastic resonance could induce oscillations. This comparative analysis of deterministic and stochastic dynamics not only provides insights into the possible dynamics of CREB regulatory motifs, but also demonstrates a framework for understanding other regulatory processes with similar network architecture.

Paleoclimate data-model comparison and the role of climate forcings over the past 1500 years

The past 1500 years provide a valuable opportunity to study the response of the climate system to external forcings. However, the integration of paleoclimate proxies with climate modeling is critical to improving the understanding of climate dynamics. In this paper, a climate system model and proxy records are therefore used to study the role of natural and anthropogenic forcings in driving the global climate. The inverse and forward approaches to paleoclimate data–model comparison are applied, and sources of uncertainty are identified and discussed. In the first of two case studies, the climate model simulations are compared with multiproxy temperature reconstructions. Robust solar and volcanic signals are detected in Southern Hemisphere temperatures, with a possible volcanic signal detected in the Northern Hemisphere. The anthropogenic signal dominates during the industrial period. It is also found that seasonal and geographical biases may cause multiproxy reconstructions to overestimate the magnitude of the long-term preindustrial cooling trend. In the second case study, the model simulations are compared with a coral δ18O record from the central Pacific Ocean. It is found that greenhouse gases, solar irradiance, and volcanic eruptions all influence the mean state of the central Pacific, but there is no evidence that natural or anthropogenic forcings have any systematic impact on El Niño–Southern Oscillation. The proxy climate relationship is found to change over time, challenging the assumption of stationarity that underlies the interpretation of paleoclimate proxies. These case studies demonstrate the value of paleoclimate data–model comparison but also highlight the limitations of current techniques and demonstrate the need to develop alternative approaches.