Across scientific disciplines, there is a rapidly growing recognition of the need for more statistically robust, transparent approaches to data visualization. Complementary to this, many scientists have called for plotting tools that accurately and transparently convey key aspects of statistical effects and raw data with minimal distortion. Previously common approaches, such as plotting conditional mean or median barplots together with error-bars have been criticized for distorting effect size, hiding underlying patterns in the raw data, and obscuring the assumptions upon which the most commonly used statistical tests are based. Here we describe a data visualization approach which overcomes these issues, providing maximal statistical information while preserving the desired ‘inference at a glance’ nature of barplots and other similar visualization devices. These “raincloud plots” can visualize raw data, probability density, and key summary statistics such as median, mean, and relevant confidence intervals in an appealing and flexible format with minimal redundancy. In this tutorial paper, we provide basic demonstrations of the strength of raincloud plots and similar approaches, outline potential modifications for their optimal use, and provide open-source code for their streamlined implementation in R, Python and Matlab (https://github.com/RainCloudPlots/RainCloudPlots). Readers can investigate the R and Python tutorials interactively in the browser using Binder by Project Jupyter.

Across scientific disciplines, there is a rapidly growing recognition of the need for more statistically robust, transparent approaches to data visualization. Complementary to this, many scientists have called for plotting tools that accurately and transparently convey key aspects of statistical effects and raw data with minimal distortion. Previously common approaches, such as plotting conditional mean or median barplots together with error-bars have been criticized for distorting effect size, hiding underlying patterns in the raw data, and obscuring the assumptions upon which the most commonly used statistical tests are based. Here we describe a data visualization approach which overcomes these issues, providing maximal statistical information while preserving the desired ‘inference at a glance’ nature of barplots and other similar visualization devices. These “raincloud plots” can visualize raw data, probability density, and key summary statistics such as median, mean, and relevant confidence intervals in an appealing and flexible format with minimal redundancy. In this tutorial paper, we provide basic demonstrations of the strength of raincloud plots and similar approaches, outline potential modifications for their optimal use, and provide open-source code for their streamlined implementation in R, Python and Matlab (https://github.com/RainCloudPlots/RainCloudPlots). Readers can investigate the R and Python tutorials interactively in the browser using Binder by Project Jupyter.

Double inversion recovery (DIR) detects only a minority (<20%) of cortical lesions (CL) in multiple sclerosis (MS). Phase-sensitive inversion recovery (PSIR) was suggested to be substantially superior to DIR in the detection of cortical lesions (CL). These two sequences might be complementary. To analyze CL frequency and type in MS patients having different disease duration and disability, including patients at clinical onset, and to discern more correctly the artifacts, by combining DIR and PSIR images. 40 patients were enrolled in the study: 10 clinically isolated syndrome/early relapsing remitting MS (CIS/eRRMS), 24 relapsing remitting MS (RRMS), 6 secondary progressive MS (SPMS). DIR and PSIR images were jointly used to classify lesions as purely intracortical (IC), leukocortical (LC) and juxtacortical (JC). PSIR disclosed CL in 100% of the patients and was capable of identifying more than four times lesions (455.5%, p<0.00001), especially IC (mean numbers: 36.5 in CIS/eRRMS, 45.0 in RRMS and 52.3 in SPMS) and LC (mean numbers: 10.9 in CIS/eRRMS, 20.1 in RRMS and 25.3 in SPMS), compared to DIR (p<0.00001). CL number was significantly higher in SPMS compared to RRMS (p<0.0001). Artifacts were more accurately identified by comparing the two sequences. Our study confirms the higher ability of PSIR in disclosing and classifying CL. The presence of CL in all CIS patients further points out the relevance of cortical pathology in MS. Whether the parallel analysis of DIR and PSIR images may be useful for diagnostic purposes, especially when a diagnosis of MS is suspected but not confirmed by routine MRI, needs to be evaluated in larger patient series. The analysis of the cortex by DIR and PSIR may also allow a better stratification of the patients for prognostic and counseling purposes, as well as for their inclusion in clinical studies.

Across scientific disciplines, there is a rapidly growing recognition of the need for more statistically robust, transparent approaches to data visualization. Complimentary to this, many scientists have realized the need for plotting tools that accurately and transparently convey key aspects of statistical effects and raw data with minimal distortion. Previously common approaches, such as plotting conditional mean or median barplots together with error-bars have been criticized for distorting effect size, hiding underlying patterns in the raw data, and obscuring the assumptions upon which the most commonly used statistical tests are based. Here we describe a data visualization approach which overcomes these issues, providing maximal statistical information while preserving the desired ‘inference at a glance’ nature of barplots and other similar visualization devices. These “raincloud plots” can visualize raw data, probability density, and key summary statistics such as median, mean, and relevant confidence intervals in an appealing and flexible format with minimal redundancy. In this tutorial paper we provide basic demonstrations of the strength of raincloud plots and similar approaches, outline potential modifications for their optimal use, and provide open-source code for their streamlined implementation in R, Python and Matlab ( https://github.com/RainCloudPlots/RainCloudPlots ). Readers can investigate the R and Python tutorials interactively in the browser using Binder by Project Jupyter.

B lymphocytes are thought to play a relevant role in multiple sclerosis (MS) pathology. The in vivo analysis of intrathecally produced B cell-related cytokines may help to clarify the mechanisms of B cell recruitment and immunoglobulin production within the central nervous system (CNS) in MS. Paired cerebrospinal fluid (CSF) and serum specimens from 40 clinically isolated syndrome suggestive of MS or early-onset relapsing-remitting MS patients (CIS/eRRMS) and 17 healthy controls (HC) were analyzed for the intrathecal synthesis of IgG (quantitative formulae and IgG oligoclonal bands, IgGOB), CXCL13, BAFF, and IL-21. 3D-FLAIR, 3D-DIR, and 3D-T1 MRI sequences were applied to evaluate white matter (WM) and gray matter (GM) lesions and global cortical thickness (gCTh). Compared to HC, CIS/eRRMS having IgGOB (IgGOB+, 26 patients) had higher intrathecal IgG indexes (p < 0.01), lower values of BAFF Index (11.9 ± 6.1 vs 17.5 ± 5.2, p < 0.01), and higher CSF CXCL13 levels (27.7 ± 33.5 vs 0.9 ± 1.5, p < 0.005). In these patients, BAFF Index but not CSF CXCL13 levels inversely correlated with the intrathecal IgG synthesis (r > 0.5 and p < 0.05 for all correlations). CSF leukocyte counts were significantly higher in IgGOB+ compared to IgGOB− (p < 0.05) and HC (p < 0.01), and correlated to CSF CXCL13 concentrations (r 0.77, p < 0.001). The gCTh was significantly lower in patients with higher CSF CXCL13 levels (2.41 ± 0.1 vs 2.49 ± 0.1 mm, p < 0.05), while no difference in MRI parameters of WM and GM pathology was observed between IgGOB+ and IgGOB−. The intrathecal IgG synthesis inversely correlated with BAFF Index and showed no correlation with CSF CXCL13. These findings seem to indicate that intrathecally synthesized IgG are produced by long-term PCs that have entered the CNS from the peripheral blood, rather than produced by PCs developed in the meningeal follicle-like structures (FLS). In this study, CXCL13 identifies a subgroup of MS patients characterized by higher leukocyte counts in the CSF and early evidence of cortical thinning, further suggesting a role for this chemokine as a possible marker of disease severity.