Benchmarking atlas-level data integration in single-cell genomics

Single-cell atlases often include samples that span locations, laboratories and conditions, leading to complex, nested batch effects in data. Thus, joint analysis of atlas datasets requires reliable data integration. To guide integration method choice, we benchmarked 68 method and preprocessing combinations on 85 batches of gene expression, chromatin accessibility and simulation data from 23 publications, altogether representing >1.2 million cells distributed in 13 atlas-level integration tasks. We evaluated methods according to scalability, usability and their ability to remove batch effects while retaining biological variation using 14 evaluation metrics. We show that highly variable gene selection improves the performance of data integration methods, whereas scaling pushes methods to prioritize batch removal over conservation of biological variation. Overall, scANVI, Scanorama, scVI and scGen perform well, particularly on complex integration tasks, while single-cell ATAC-sequencing integration performance is strongly affected by choice of feature space. Our freely available Python module and benchmarking pipeline can identify optimal data integration methods for new data, benchmark new methods and improve method development.

Experimental animal myopia models are applicable to human juvenile-onset myopia

Landmark explorations by Hubel and Wiesel investigating the importance of visual impressions in postnatal development of the visual system demonstrated that neural connections and eye growth can be affected by the absence of a clear retinal image during a critical period of postnatal development. Fundamental theories on neural plasticity and deprivation have recently been established that presume that a reduced quality of the retinal image during infancy and early childhood triggers an elongation of the posterior chamber of the eye, a so-called form deprivation myopia (FDM). In a retrospective multicenter study of 187 patients who suffered from phlyctenular keratitis with corneal opacification since early childhood, we reviewed data on gender, year and age at onset of the disease, refraction, and ultrasound biometry. Compared with the average refraction of +0.5 diopter (D) found in the general population, the mean refraction of -4.43 D that we found in our study demonstrated a marked shift toward myopia of almost 5 D. Patients with an early onset of phlyctenular keratitis had considerably higher myopia (-6.68 D) than those with a late onset (-1.67 D). Additionally, an axial elongation was confirmed by ultrasound biometry. Our average, axial length was 26.53 mm, compared with the epidemiologic mean of 24.00 mm. This myopic shift of 2.53 mm was caused mainly by an enlarged vitreous cavity. These results support the finding that blur can affect eye growth and lead to FDM not only in animal experiments but also in human beings.

Impact of CT on diagnosis and management of acute abdomen in patients initially treated without surgery

OBJECTIVE

The purpose of this study was to evaluate the effect of CT on the diagnosis and management of acute abdominal pain in patients who did not undergo surgery and to determine what population of patients would profit most from CT examination.

MATERIALS AND METHODS

Clinical data and CT reports of 91 patients with acute abdomen (41 men and 50 women, 22-96 years old) were analyzed retrospectively. The accuracies of clinical evaluation and CT in revealing the cause of acute abdomen were compared, and the effect of CT on patient management was assessed. Analysis included the entire population of patients and these subgroups: (1) patients who had symptoms for fewer than 24 hr versus patients who had symptoms for 24 hr or more and (2) patients who had a history of abdominal diseases versus patients who had no such history.

RESULTS

Twenty-nine patients had signs or symptoms for fewer than 24 hr, and 62 patients had signs or symptoms for 24 hr or more. Fifty-nine patients had a history of abdominal disease, and 32 had no history of abdominal disease. In the entire population of patients, CT was superior to clinical evaluation for diagnosing the cause of acute abdomen (sensitivity was 90% for CT and 76% for clinical evaluation, p < .0005). Management was changed after CT in 25 patients (p < .0005). Similar differences were observed in the subgroups of patients with signs and symptoms for fewer than 24 hr, patients with signs and symptoms for 24 hr or more, and patients with no history of abdominal disease (p < .05). In the subgroup of patients with a history of abdominal disease, the differences between clinical evaluation and CT were not statistically significant.

CONCLUSION

CT is an excellent examination technique for patients with acute abdomen, regardless of the duration of signs and symptoms. CT is particularly useful in defining the cause and therapeutic strategy in patients with acute abdomen who have no history of abdominal disease.

Brain lesions in patients with multiple sclerosis: detection with echo-planar imaging

PURPOSE

To evaluate the detection of brain lesions with echo-planar imaging relative to conventional spin-echo (SE) imaging.

MATERIALS AND METHODS

In 17 patients (three men, 14 women; mean age, 31 years) with multiple sclerosis, the following were compared: single-shot proton-density- and T2-weighted and thin-section T2-weighted echo-planar, proton-density- and T2-weighted multishot echo-planar, and conventional SE sequences. Quantitative and qualitative criteria as well as lesion detectability were evaluated. The proton-density-weighted SE sequence was used as the standard of reference.

RESULTS

Multishot sequences were superior to single-shot sequences in image quality and lesion detectability. With the multishot proton-density-weighted sequence, 53 of 54 large lesions and 23 of 30 small lesions were detected; with the single-shot proton-density-weighted sequence, 38 of 54 large lesions and five of 30 small lesions were detected.

CONCLUSION

With multishot echo-planar sequences, detectability of large lesions is similar to that with conventional SE imaging. Susceptibility artifact is diminished in comparison to single-shot echo-planar sequences.