An end-to-end framework for private DGA detection as a service.

Domain Generation Algorithms (DGAs) are used by malware to generate pseudorandom domain names to establish communication between infected bots and command and control servers. While DGAs can be detected by machine learning (ML) models with great accuracy, offering DGA detection as a service raises privacy concerns when requiring network administrators to disclose their DNS traffic to the service provider. The main scientific contribution of this paper is to propose the first end-to-end framework for privacy-preserving classification as a service of domain names into DGA (malicious) or non-DGA (benign) domains. Our framework achieves these goals by carefully designed protocols that combine two privacy-enhancing technologies (PETs), namely secure multi-party computation (MPC) and differential privacy (DP). Through MPC, our framework enables an enterprise network administrator to outsource the problem of classifying a DNS (Domain Name System) domain as DGA or non-DGA to an external organization without revealing any information about the domain name. Moreover, the service provider's ML model used for DGA detection is never revealed to the network administrator. Furthermore, by using DP, we also ensure that the classification result cannot be used to learn information about individual entries of the training data. Finally, we leverage post-training float16 quantization of deep learning models in MPC to achieve efficient, secure DGA detection. We demonstrate that by using quantization achieves a significant speed-up, resulting in a 23% to 42% reduction in inference runtime without reducing accuracy using a three party secure computation protocol tolerating one corruption. Previous solutions are not end-to-end private, do not provide differential privacy guarantees for the model's outputs, and assume that model embeddings are publicly known. Our best protocol in terms of accuracy runs in about 0.22s.

High performance logistic regression for privacy-preserving genome analysis.

BACKGROUND: In biomedical applications, valuable data is often split between owners who cannot openly share the data because of privacy regulations and concerns. Training machine learning models on the joint data without violating privacy is a major technology challenge that can be addressed by combining techniques from machine learning and cryptography. When collaboratively training machine learning models with the cryptographic technique named secure multi-party computation, the price paid for keeping the data of the owners private is an increase in computational cost and runtime. A careful choice of machine learning techniques, algorithmic and implementation optimizations are a necessity to enable practical secure machine learning over distributed data sets. Such optimizations can be tailored to the kind of data and Machine Learning problem at hand. METHODS: Our setup involves secure two-party computation protocols, along with a trusted initializer that distributes correlated randomness to the two computing parties. We use a gradient descent based algorithm for training a logistic regression like model with a clipped ReLu activation function, and we break down the algorithm into corresponding cryptographic protocols. Our main contributions are a new protocol for computing the activation function that requires neither secure comparison protocols nor Yao's garbled circuits, and a series of cryptographic engineering optimizations to improve the performance. RESULTS: For our largest gene expression data set, we train a model that requires over 7 billion secure multiplications; the training completes in about 26.90 s in a local area network. The implementation in this work is a further optimized version of the implementation with which we won first place in Track 4 of the iDASH 2019 secure genome analysis competition. CONCLUSIONS: In this paper, we present a secure logistic regression training protocol and its implementation, with a new subprotocol to securely compute the activation function. To the best of our knowledge, we present the fastest existing secure multi-party computation implementation for training logistic regression models on high dimensional genome data distributed across a local area network.

On the Composability of Statistically Secure Random Oblivious Transfer.

We show that random oblivious transfer protocols that are statistically secure according to a definition based on a list of information-theoretical properties are also statistically universally composable. That is, they are simulatable secure with an unlimited adversary, an unlimited simulator, and an unlimited environment machine. Our result implies that several previous oblivious transfer protocols in the literature that were proven secure under weaker, non-composable definitions of security can actually be used in arbitrary statistically secure applications without lowering the security.

Essential oils for managing Pratylenchus penetrans on Easter lilies.

Easter lily bulbs for greenhouse forcing are produced in Del Norte County, California and Curry County, Oregon, USA. Pratylenchus penetrans infestation seriously affects growth of field grown bulbs. During two consecutive years of field trials containing 22 treatments, commercially prepared formulations of essential oils (EOs) were compared to an untreated control and to a standard chemical fumigant treatment (FU) (1,3-dichloropropene and metam sodium) applied preplant followed by phorate (PH) at planting to determine their value in the management of lesion nematode, and in improving plant health. The EO products Duogard, EF400, EF300, and Cinnamite were tested as preplant dips to bulblet planting stock. The treated bulblets were tested either alone, in combination with PH at-planting, at planting following FU or in combination with PH at planting following FU. The organophosphates ethoprop and fosthiazate were also tested either alone, or at a reduced rate in combination with a reduced rate of PH. With respect to bulb circumference, ten treatments consistently outperformed the control. In consecutive years, three treatments had healthier looking roots than the control. At harvest, levels of lesion nematode within roots were consistently lower in nine treatments. EOs were beneficial in mitigating nematode damage.Easter lily bulbs for greenhouse forcing are produced in Del Norte County, California and Curry County, Oregon, USA. Pratylenchus penetrans infestation seriously affects growth of field grown bulbs. During two consecutive years of field trials containing 22 treatments, commercially prepared formulations of essential oils (EOs) were compared to an untreated control and to a standard chemical fumigant treatment (FU) (1,3-dichloropropene and metam sodium) applied preplant followed by phorate (PH) at planting to determine their value in the management of lesion nematode, and in improving plant health. The EO products Duogard, EF400, EF300, and Cinnamite were tested as preplant dips to bulblet planting stock. The treated bulblets were tested either alone, in combination with PH at-planting, at planting following FU or in combination with PH at planting following FU. The organophosphates ethoprop and fosthiazate were also tested either alone, or at a reduced rate in combination with a reduced rate of PH. With respect to bulb circumference, ten treatments consistently outperformed the control. In consecutive years, three treatments had healthier looking roots than the control. At harvest, levels of lesion nematode within roots were consistently lower in nine treatments. EOs were beneficial in mitigating nematode damage.

Protecting Privacy of Users in Brain-Computer Interface Applications.

Machine learning (ML) is revolutionizing research and industry. Many ML applications rely on the use of large amounts of personal data for training and inference. Among the most intimate exploited data sources is electroencephalogram (EEG) data, a kind of data that is so rich with information that application developers can easily gain knowledge beyond the professed scope from unprotected EEG signals, including passwords, ATM PINs, and other intimate data. The challenge we address is how to engage in meaningful ML with EEG data while protecting the privacy of users. Hence, we propose cryptographic protocols based on secure multiparty computation (SMC) to perform linear regression over EEG signals from many users in a fully privacy-preserving (PP) fashion, i.e., such that each individual's EEG signals are not revealed to anyone else. To illustrate the potential of our secure framework, we show how it allows estimating the drowsiness of drivers from their EEG signals as would be possible in the unencrypted case, and at a very reasonable computational cost. Our solution is the first application of commodity-based SMC to EEG data, as well as the largest documented experiment of secret sharing-based SMC in general, namely, with 15 players involved in all the computations.

Correlates of T50 and relationships with bone mineral density in community-living older men: the osteoporotic fractures in men (MrOS) study.

PURPOSE: T50 is a novel serum-based marker that assesses the propensity of calcification in serum. Shorter T50 indicates greater propensity to calcify and it has been associated to cardiovascular disease (CVD) and mortality among patients with kidney disease. In the general population, neither the correlates of T50 nor the relationships of T50 with bone mineral density (BMD) are known. METHODS: We performed a nested cross-sectional study selecting 150 individuals at random among participants from the Osteoporotic Fractures in Men (MrOS) Study, a study of community-living older men. We categorized individuals into tertiles of T50 and compared demographics and disease indicators across tertiles. We utilized linear regression to evaluate the cross-sectional association between T50 and hip and spine BMD in multivariable models. RESULTS: Older age was associated with shorter T50. Kidney function tended to be lower in those with shorter T50 and the prevalence of CVD and peripheral arterial disease in those with shorter T50, albeit these findings did not achieve statistical significance. We found no statistically significant associations between T50 and total hip or total spine BMD in either unadjusted or multivariable adjusted models. CONCLUSIONS: T50, a novel indicator of serum calcification propensity, is not associated with BMD in community-living older men. Future larger studies should determine if T50 may give insights to CVD in the general population above and beyond traditional risk factors.

NOX1 loss-of-function genetic variants in patients with inflammatory bowel disease.

Genetic defects that affect intestinal epithelial barrier function can present with very early-onset inflammatory bowel disease (VEOIBD). Using whole-genome sequencing, a novel hemizygous defect in NOX1 encoding NAPDH oxidase 1 was identified in a patient with ulcerative colitis-like VEOIBD. Exome screening of 1,878 pediatric patients identified further seven male inflammatory bowel disease (IBD) patients with rare NOX1 mutations. Loss-of-function was validated in p.N122H and p.T497A, and to a lesser degree in p.Y470H, p.R287Q, p.I67M, p.Q293R as well as the previously described p.P330S, and the common NOX1 SNP p.D360N (rs34688635) variant. The missense mutation p.N122H abrogated reactive oxygen species (ROS) production in cell lines, ex vivo colonic explants, and patient-derived colonic organoid cultures. Within colonic crypts, NOX1 constitutively generates a high level of ROS in the crypt lumen. Analysis of 9,513 controls and 11,140 IBD patients of non-Jewish European ancestry did not reveal an association between p.D360N and IBD. Our data suggest that loss-of-function variants in NOX1 do not cause a Mendelian disorder of high penetrance but are a context-specific modifier. Our results implicate that variants in NOX1 change brush border ROS within colonic crypts at the interface between the epithelium and luminal microbes.

Operative and survival outcomes in a series of 100 consecutive cases of robot-assisted transhiatal esophagectomies.

Robotic-assisted transhiatal esophagectomy (RATE) is a technically complex procedure with potential for improved postoperative outcomes. In this report, we describe our experience with RATE in a large case series. A retrospective review was conducted to collect clinical, outcomes, and survival data for 100 consecutive patients with esophageal cancer (n = 98) and benign (n = 2) conditions undergoing RATE between March 2007 and December 2014. Progression-free (PFS) and overall (OS) survival were estimated using the Kaplan-Meier curves with comparisons by log-rank tests. Median operative time and estimated blood loss were 264 minutes and 75 mL, respectively. Median intensive care unit stay was 1 day and median length of hospital stay was 8 days. Postoperative complications commonly observed were nonmalignant pleural effusion (38%) and recurrent laryngeal nerve injury (33%); 30 day mortality rate was 2%. Median number of lymph nodes removed during RATE was 17 and R0 resection was achieved in 97.8% patients. At the end of the median follow-up period of 27.7 months, median PFS was 41 months and median OS was 54 months. 1-year and 3-year PFS rates were 82% (95% CI, 75%-89%) and 53% (95% CI, 42%-62%), respectively, and OS rates were 95% (95% CI, 91%-99%) and 57% (95% CI, 46%-67%). In our experience, RATE is an effective and safe oncologic surgical procedure in a carefully selected group of patients with acceptable operative time, minimal blood loss, standard postoperative morbidity and adequate PFS and OS profiles.

Blending process modeling and control by multivariate curve resolution.

The application of the Multivariate Curve Resolution by Alternating Least Squares (MCR-ALS) method to model and control blend processes of pharmaceutical formulations is assessed. Within the MCR-ALS framework, different data analysis approaches have been tested depending on the objective of the study, i.e., knowing the effect of different factors in the evolution of the blending process (modeling) or detecting the blend end-point and monitoring the concentration of the different species during and at the end of the process (control). Data analysis has been carried out studying multiple blending runs simultaneously taking advantage of the multiset mode of the MCR-ALS method. During the ALS optimization, natural constraints, such as non-negativity (spectral and concentration directions) have been applied for blend modeling. When blending control is the main purpose, a variant of the MCR-ALS algorithm with correlation constraint in the concentration direction has been additionally used. This constraint incorporates an internal calibration procedure, which relates resolved concentration values (in arbitrary units) with the real reference concentration values in the calibration samples (known references) providing values in real concentration scale in the final MCR-ALS results. Two systems consisting of pharmaceutical mixtures of an active principle (acetaminophen) with two or four excipients have been investigated. In the first case, MCR results allowed the description of the evolution of the individual compounds and the assessment of some physical effects in the blending process. In the second case, MCR analysis allowed the detection of the end-point of the process and the assessment of the effects linked to variations in the concentration level of the compounds.

optiCall: a robust genotype-calling algorithm for rare, low-frequency and common variants.

MOTIVATION: Existing microarray genotype-calling algorithms adopt either SNP-by-SNP (SNP-wise) or sample-by-sample (sample-wise) approaches to calling. We have developed a novel genotype-calling algorithm for the Illumina platform, optiCall, that uses both SNP-wise and sample-wise calling to more accurately ascertain genotypes at rare, low-frequency and common variants. RESULTS: Using data from 4537 individuals from the 1958 British Birth Cohort genotyped on the Immunochip, we estimate the proportion of SNPs lost to downstream analysis due to false quality control failures, and rare variants misclassified as monomorphic, is only 1.38% with optiCall, in comparison to 3.87, 7.85 and 4.09% for Illuminus, GenoSNP and GenCall, respectively. We show that optiCall accurately captures rare variants and can correctly account for SNPs where probe intensity clouds are shifted from their expected positions. AVAILABILITY AND IMPLEMENTATION: optiCall is implemented in C++ for use on UNIX operating systems and is available for download at http://www.sanger.ac.uk/resources/software/opticall/.

Potassium intake and risk of incident type 2 diabetes mellitus: the Coronary Artery Risk Development in Young Adults (CARDIA) Study.

AIMS/HYPOTHESIS: Serum potassium has been found to be a significant predictor of diabetes risk, but the effect of dietary potassium on diabetes risk is not clear. We sought to determine if dietary potassium is associated with risk of incident type 2 diabetes in young adults. METHODS: We used data from the Coronary Artery Risk Development in Young Adults (CARDIA) Study. Potassium intake was measured by (1) an average of three 24 h urinary potassium collections at the 5-year study visit, and (2) the CARDIA dietary assessment instrument at baseline. Incident type 2 diabetes cases were ascertained on the basis of use of diabetes medication and laboratory measurements. Analyses were adjusted for relevant confounders including intake of fruit and vegetables and other dietary factors. RESULTS: Of 1,066 participants with urinary potassium measurements, 99 (9.3%) developed diabetes over 15 years of follow-up. In multivariate models, adults in the lowest urinary potassium quintile were more than twice as likely to develop diabetes as their counterparts in the highest quintile (HR 2.45; 95% CI 1.08, 5.59). Of 4,754 participants with dietary history measurements, 373 (7.8%) developed diabetes over 20 years of follow-up. In multivariate models, African-Americans had a significantly increased risk of diabetes with lower potassium intake, which was not found in whites. CONCLUSIONS/INTERPRETATION: Low dietary potassium is associated with increased risk of incident diabetes in African-Americans. Randomised clinical trials are needed to determine if potassium supplementation, from either dietary or pharmacological sources, could reduce the risk of diabetes, particularly in higher-risk populations.

Sucrose and dental caries: a review of the evidence.

The aim of this study was to conduct a review of the literature to assess the relationship between quantity and pattern of sucrose use and dental caries. Using hand and electronic methods (MEDLINE, EMBASE) the literature was searched for epidemiological papers concerning any relationship of sugars and dental caries published since 1856. Superficial hand searching was carried out between 1856 and 1940, detailed hand searching 1940-1966 and electronic 1966-2007. Selection criteria were set based on, but not confined to, Cochran style standards. Investigations were categorized as A, fulfilling all criteria; B1, relevant fulfilling 19 of 23 criteria; B2, relevant but fulfilling only between 12 and 18 of the selection criteria; and C, all other papers. There were 95 papers meeting most (more than 12) or all of the selected criteria. Only 1 paper was graded A; 31 as B1. There were in addition some 65 as B2 and all the rest as C, which were discarded. There were a wide variety of study designs and those graded A or B1 comprised 23 ecological cross-sectional, 7 cohort and 2 case control studies. Summary results showed that 6 papers found a positive, significant relationship of sugar quantity to dental caries, 19 of 31 studies reported a significant relationship of sugar frequency of use to dental caries. The balance of studies does not demonstrate a relationship between sugar quantity, but a moderately significant relationship of sugar frequency to dental caries.

Tumorigenicity assessments of Per.C6 cells and of an Ad5-vectored HIV-1 vaccine produced on this continuous cell line.

PER.C6, a cell line derived from human embryonic retinal cells transformed with the Adenovirus Type 5 (Ad5) E1A and E1B genes, is used to produce E1-deleted Ad5 vectors such as the MRKAd5 HIV-1 gag vaccine. While whole, live PER.C6 cells are capable of growing as tumours when transplanted subcutaneously into immunodeficient nude mice at a high dosage, the process for vaccine production includes filtration steps and other methods which effectively preclude contamination by intact viable substrate cells. However, because of the neoplastic nature of this cell line, we carried out a series of investigations to assess the tumorigenic risk posed by residuals from the cell substrate in a vaccine. To address concerns about transmission of oncogenic DNA, we demonstrated that purified PER.C6 cellular DNA does not induce tumours in newborn hamsters or nude mice. To address concerns about other potential residuals, including hypothetical adventitious tumour viruses, we demonstrated that a PER.C6 cell lysate and a MRKAd5 HIV-1 gag vaccine produced on PER.C6 cells do not induce tumours in newborn hamsters or newborn rats. These results, in conjunction with the wide panel of viral safety tests performed on these cells, support the safety of the PER.C6 as a cell substrate for vaccine production.

Anal condyloma: a comparison between HIV positive and negative patients.

HIV positive and negative patients with anal condylomata were compared to determine an association with squamous cell neoplasia, its disease progression, and response to treatment. From 1992 to 2003, 61 patients were diagnosed with anal condylomata by anal biopsy. Thirty-four patients were HIV+ and 27 patients were considered HIV-. Upon retrospective chart review, details on disease progression, development of malignancy, and subsequent treatment were collected. Sixty-one per cent of HIV+ patients had a neoplastic process in contrast to 25 per cent of HIV- patients (P = 0.005). Five patients demonstrated disease progression, of which four were HIV+. Three HIV+ patients were treated for invasive carcinoma with excision and standard chemoradiation therapy. Two patients with T3 lesions developed recurrence and died. Eighteen HIV+ patients had noninvasive carcinoma and were treated with local excision without recurrence at mean follow-up of 28 months. HIV+ patients were shown to have more condylomata harboring squamous cell neoplasia than HIV- patients. Noninvasive carcinoma can be treated effectively with local excision, independent of HIV status; however, long-term follow-up is needed. Chemoradiation therapy in patients who are relatively healthy and have stage I disease may be successful. The role for chemoradiation in AIDS patients with stage III disease remains unclear.

Automated distal coronary bypass with a novel magnetic coupler (MVP system).

OBJECTIVE: We sought to assess the feasibility of performing sutureless distal coronary artery bypass anastomoses with a novel magnetic coupling device. METHODS: From May 2000 to April 2001, single-vessel side-to-side coronary artery bypass grafting on a beating heart was performed in 39 domestic white pigs (35-60 kg) without the use of mechanical stabilization, shunts, or perfusion bridges. Animals were divided into 2 groups. Seventeen pigs underwent right internal thoracic artery to right coronary artery bypass grafting through a median sternotomy (group 1) with a novel magnetic vascular positioning system (MVP system; Ventrica, Inc, Fremont, Calif). Twenty-two pigs underwent left internal thoracic artery to left anterior descending artery grafting with the MVP anastomotic device through a left anterior minithoracotomy (group 2). This system consists of 2 pairs of elliptical magnetic implants and a deployment device. One pair of magnets forms the anastomotic docking port within the graft; the other pair forms an identical anastomotic docking port within the target vessel. The anastomosis is created when the 2 docking ports magnetically couple. Anastomotic patency was evaluated by means of angiography during the first postoperative week and at 1 month. Histologic studies were performed at different time points as late as 6 months. RESULTS: Right internal thoracic artery to right coronary artery anastomoses and left internal thoracic artery to left anterior descending artery anastomoses were successfully performed with the system in all animals. The self-adherent and self-aligning properties of the implants allowed for immediate and secure approximation of the arteries (total anastomotic time between 2-3 minutes). Anastomoses were constructed without a stabilization platform. Five nondevice-related deaths occurred postoperatively. One-week angiography, performed in 35 surviving animals, showed a patent graft and anastomosis in all cases. The patency rate at 1 month was 97% (33/34). Histologic studies as late as 6 months demonstrated neointimal coverage of the magnets without any significant luminal obstruction. Histology also confirmed the presence of viable tissue between magnets. CONCLUSION: The MVP anastomotic system uses magnetic force to create rapid and secure distal coronary artery anastomoses, which might facilitate minimally invasive and totally endoscopic coronary artery bypass surgery.

Management Options for Pratylenchus penetrans in Easter Lily.

Alternatives to reduce or modify nematicide use for minimizing groundwater contamination in Easter lily were explored in two field trials. Alternatives to standard 1,3-dichloropropene (1,3-D) plus phorate injection in the first trial were: (i) delaying applications until after winter rains, (ii) removing roots from planting stock, (iii) 1,3-D via drip irrigation, (iv) a chitin-urea soil amendment, (v) the registered insecticide disulfoton, and (vi) several nonregistered nematicides. None of the treatments equaled the standard treatment. In the second trial, potential benefits of adding a systemic nematicide, oxamyl (OX), or a fungicide, metalaxyl (MX), to the standard treatment were explored. Preplant drip irrigation applications of metam sodium (MS), sodium tetrathiocarbonate (ST), and emulsifiable 1,3-D were evaluated alone and in combination with postplant applications of OX and MX. Several drip-applied treatments performed comparably to the standard treatment with respect to the most important criteria of crop quality, bulb circumference. Metam-sodium in combination with either or both OX and MX, 1,3-D plus OX and MX, and ST plus OX and MX provided the best results.

A response of Panc-1 cells to cis-platinum, assessed with a cDNA array.

BACKGROUND: The problem posed by the lack of response of cells in most solid cancers to current chemotherapy generally remains intractable. MATERIALS AND METHODS: The use of cDNA arrays represents one global approach to identifying reasons for this failure. A messenger RNA response of pancreatic cancer (Panc-1) cells after culture for 24 hours with 12 microM cis-platinum was analyzed with a commercial cDNA array. RESULTS: Major drug-induced events included inhibition of messenger RNAs associated with cell proliferation and up-regulation of generally countervailing DNA repair, cellular stress, heat shock protein, glutathione stress-related and multiple drug resistance enzyme messenger RNAs, accompanied by a limited programmed cell death response. CONCLUSION: Induction of widespread normal stress-induced countervailing mRNAs by comparatively non-selective agents such as cis-platinum strongly biases against a successful therapeutic outcome. This paradoxical result of a therapeutic intent provides a further compelling argument for the use of specifically-targeted therapy such as growth factor receptor, tyrosine kinase and other discretely focused agents, probably employed in combinations based on expression of their targets in an individual patient's cancer, as identified by cDNA or proteonomic arrays.

Results of laparoscopically assisted colon resection for carcinoma.

BACKGROUND: Surgical resection is the primary treatment for colorectal carcinoma. Laparoscopically assisted colon resection technically is feasible for both benign and malignant disease. However, the role of laparoscopically assisted colon resection for carcinoma is controversial. METHODS: We prospectively studied our first 100 patients with colorectal carcinoma who successfully underwent laparoscopically assisted colon resection for the carcinoma. RESULTS: The pathologic stages were Dukes' categories A-16, B-52, C-25, and D-7. Operative mortality and morbidity were 2% and 22%, respectively. During a mean follow-up period of 40.3 months, recurrence by stage was zero patients with stage A disease, five patients with stage B disease, nine patients with stage C disease. Thirteen of these patients died as a result of their disease. At this writing, 60 patients are alive without evidence of disease, and 23 have completed the study disease free after more than 60 months. The 5-year survival probabilities by stage were 100% for stage A, 76.8% for stage B, and 51.7% for stage C. CONCLUSIONS: Laparoscopically assisted colectomy for cancer can be performed safely. The recurrence rate after laparoscopically assisted resection appears to be at least as good as after open resection. Results from ongoing prospective, randomized trials are needed to confirm these findings.

A dissociation between symbolic number knowledge and analogue magnitude information.

Semantic understanding of numbers and related concepts can be dissociated from rote knowledge of arithmetic facts. However, distinctions among different kinds of semantic representations related to numbers have not been fully explored. Working with numbers and arithmetic requires representing semantic information that is both analogue (e.g., the approximate magnitude of a number) and symbolic (e.g., what / means). In this article, the authors describe a patient (MC) who exhibits a dissociation between tasks that require symbolic number knowledge (e.g., knowledge of arithmetic symbols including numbers, knowledge of concepts related to numbers such as rounding) and tasks that require an analogue magnitude representation (e.g., comparing size or frequency). MC is impaired on a variety of tasks that require symbolic number knowledge, but her ability to represent and process analogue magnitude information is intact. Her deficit in symbolic number knowledge extends to a variety of concepts related to numbers (e.g., decimal points, Roman numerals, what a quartet is) but not to any other semantic categories that we have tested. These findings suggest that symbolic number knowledge is a functionally independent component of the number processing system, that it is category specific, and that it is anatomically and functionally distinct from magnitude representations.