Characterizing efficient feature selection for single-cell expression analysis.

Unsupervised feature selection is a critical step for efficient and accurate analysis of single-cell RNA-seq data. Previous benchmarks used two different criteria to compare feature selection methods: (i) proportion of ground-truth marker genes included in the selected features and (ii) accuracy of cell clustering using ground-truth cell types. Here, we systematically compare the performance of 11 feature selection methods for both criteria. We first demonstrate the discordance between these criteria and suggest using the latter. We then compare the distribution of selected genes in their means between feature selection methods. We show that lowly expressed genes exhibit seriously high coefficients of variation and are mostly excluded by high-performance methods. In particular, high-deviation- and high-expression-based methods outperform the widely used in Seurat package in clustering cells and data visualization. We further show they also enable a clear separation of the same cell type from different tissues as well as accurate estimation of cell trajectories.

Depression and suicide risk prediction models using blood-derived multi-omics data.

More than 300 million people worldwide experience depression; annually, ~800,000 people die by suicide. Unfortunately, conventional interview-based diagnosis is insufficient to accurately predict a psychiatric status. We developed machine learning models to predict depression and suicide risk using blood methylome and transcriptome data from 56 suicide attempters (SAs), 39 patients with major depressive disorder (MDD), and 87 healthy controls. Our random forest classifiers showed accuracies of 92.6% in distinguishing SAs from MDD patients, 87.3% in distinguishing MDD patients from controls, and 86.7% in distinguishing SAs from controls. We also developed regression models for predicting psychiatric scales with R2 values of 0.961 and 0.943 for Hamilton Rating Scale for Depression-17 and Scale for Suicide Ideation, respectively. Multi-omics data were used to construct psychiatric status prediction models for improved mental health treatment.

Grading system for periodontitis by analyzing levels of periodontal pathogens in saliva.

Periodontitis is an infectious disease that is associated with microorganisms that colonize the tooth surface. Clinically, periodontal condition stability reflects dynamic equilibrium between bacterial challenge and host response. Therefore, periodontal pathogen assessment can assist in the early detection of periodontitis. Here we developed a grading system called the periodontal pathogen index (PPI) by analyzing the copy numbers of multiple pathogens both in healthy and chronic periodontitis patients. We collected 170 mouthwash samples (64 periodontally healthy controls and 106 chronic periodontitis patients) and analyzed the salivary 16S rRNA levels of nine pathogens using multiplex, quantitative real-time polymerase chain reaction. Except for Aggregatibacter actinomycetemcomitans, copy numbers of all pathogens were significantly higher in chronic periodontitis patients. We classified the samples based on optimal cut-off values with maximum sensitivity and specificity from receiver operating characteristic curve analyses (AUC = 0.91, 95% CI: 0.87-0.96) into four categories of PPI: Healthy (1-40), Moderate (41-60), At Risk (61-80), and Severe (81-100). PPI scores were significantly higher in all chronic periodontitis patients than in the controls (odds ratio: 31.7, 95% CI: 13.41-61.61) and were associated with age, scaling as well as clinical characteristics including clinical attachment level and plaque index. Our PPI grading system can be clinically useful for the early assessment of pathogenic bacterial burden and follow-up monitoring after periodontitis treatment.

Integrated Molecular Characterization of Testicular Germ Cell Tumors.

We studied 137 primary testicular germ cell tumors (TGCTs) using high-dimensional assays of genomic, epigenomic, transcriptomic, and proteomic features. These tumors exhibited high aneuploidy and a paucity of somatic mutations. Somatic mutation of only three genes achieved significance-KIT, KRAS, and NRAS-exclusively in samples with seminoma components. Integrated analyses identified distinct molecular patterns that characterized the major recognized histologic subtypes of TGCT: seminoma, embryonal carcinoma, yolk sac tumor, and teratoma. Striking differences in global DNA methylation and microRNA expression between histology subtypes highlight a likely role of epigenomic processes in determining histologic fates in TGCTs. We also identified a subset of pure seminomas defined by KIT mutations, increased immune infiltration, globally demethylated DNA, and decreased KRAS copy number. We report potential biomarkers for risk stratification, such as miRNA specifically expressed in teratoma, and others with molecular diagnostic potential, such as CpH (CpA/CpC/CpT) methylation identifying embryonal carcinomas.

Comparison of carnivore, omnivore, and herbivore mammalian genomes with a new leopard assembly.

BACKGROUND: There are three main dietary groups in mammals: carnivores, omnivores, and herbivores. Currently, there is limited comparative genomics insight into the evolution of dietary specializations in mammals. Due to recent advances in sequencing technologies, we were able to perform in-depth whole genome analyses of representatives of these three dietary groups. RESULTS: We investigated the evolution of carnivory by comparing 18 representative genomes from across Mammalia with carnivorous, omnivorous, and herbivorous dietary specializations, focusing on Felidae (domestic cat, tiger, lion, cheetah, and leopard), Hominidae, and Bovidae genomes. We generated a new high-quality leopard genome assembly, as well as two wild Amur leopard whole genomes. In addition to a clear contraction in gene families for starch and sucrose metabolism, the carnivore genomes showed evidence of shared evolutionary adaptations in genes associated with diet, muscle strength, agility, and other traits responsible for successful hunting and meat consumption. Additionally, an analysis of highly conserved regions at the family level revealed molecular signatures of dietary adaptation in each of Felidae, Hominidae, and Bovidae. However, unlike carnivores, omnivores and herbivores showed fewer shared adaptive signatures, indicating that carnivores are under strong selective pressure related to diet. Finally, felids showed recent reductions in genetic diversity associated with decreased population sizes, which may be due to the inflexible nature of their strict diet, highlighting their vulnerability and critical conservation status. CONCLUSIONS: Our study provides a large-scale family level comparative genomic analysis to address genomic changes associated with dietary specialization. Our genomic analyses also provide useful resources for diet-related genetic and health research.

Integrative and comparative genomic analysis of HPV-positive and HPV-negative head and neck squamous cell carcinomas.

PURPOSE: The genetic differences between human papilloma virus (HPV)-positive and -negative head and neck squamous cell carcinomas (HNSCC) remain largely unknown. To identify differential biology and novel therapeutic targets for both entities, we determined mutations and copy-number aberrations in a large cohort of locoregionally advanced HNSCC. EXPERIMENTAL DESIGN: We performed massively parallel sequencing of 617 cancer-associated genes in 120 matched tumor/normal samples (42.5% HPV-positive). Mutations and copy-number aberrations were determined and results validated with a secondary method. RESULTS: The overall mutational burden in HPV-negative and HPV-positive HNSCC was similar with an average of 15.2 versus 14.4 somatic exonic mutations in the targeted cancer-associated genes. HPV-negative tumors showed a mutational spectrum concordant with published lung squamous cell carcinoma analyses with enrichment for mutations in TP53, CDKN2A, MLL2, CUL3, NSD1, PIK3CA, and NOTCH genes. HPV-positive tumors showed unique mutations in DDX3X, FGFR2/3 and aberrations in PIK3CA, KRAS, MLL2/3, and NOTCH1 were enriched in HPV-positive tumors. Currently targetable genomic alterations were identified in FGFR1, DDR2, EGFR, FGFR2/3, EPHA2, and PIK3CA. EGFR, CCND1, and FGFR1 amplifications occurred in HPV-negative tumors, whereas 17.6% of HPV-positive tumors harbored mutations in fibroblast growth factor receptor genes (FGFR2/3), including six recurrent FGFR3 S249C mutations. HPV-positive tumors showed a 5.8% incidence of KRAS mutations, and DNA-repair gene aberrations, including 7.8% BRCA1/2 mutations, were identified. CONCLUSIONS: The mutational makeup of HPV-positive and HPV-negative HNSCC differs significantly, including targetable genes. HNSCC harbors multiple therapeutically important genetic aberrations, including frequent aberrations in the FGFR and PI3K pathway genes. See related commentary by Krigsfeld and Chung, p. 495.

KDM4A lysine demethylase induces site-specific copy gain and rereplication of regions amplified in tumors.

Acquired chromosomal instability and copy number alterations are hallmarks of cancer. Enzymes capable of promoting site-specific copy number changes have yet to be identified. Here, we demonstrate that H3K9/36me3 lysine demethylase KDM4A/JMJD2A overexpression leads to localized copy gain of 1q12, 1q21, and Xq13.1 without global chromosome instability. KDM4A-amplified tumors have increased copy gains for these same regions. 1q12h copy gain occurs within a single cell cycle, requires S phase, and is not stable but is regenerated each cell division. Sites with increased copy number are rereplicated and have increased KDM4A, MCM, and DNA polymerase occupancy. Suv39h1/KMT1A or HP1γ overexpression suppresses the copy gain, whereas H3K9/K36 methylation interference promotes gain. Our results demonstrate that overexpression of a chromatin modifier results in site-specific copy gains. This begins to establish how copy number changes could originate during tumorigenesis and demonstrates that transient overexpression of specific chromatin modulators could promote these events.

Tandem duplication of chromosomal segments is common in ovarian and breast cancer genomes.

The application of paired-end next generation sequencing approaches has made it possible to systematically characterize rearrangements of the cancer genome to base-pair level. Utilizing this approach, we report the first detailed analysis of ovarian cancer rearrangements, comparing high-grade serous and clear cell cancers, and these histotypes with other solid cancers. Somatic rearrangements were systematically characterized in eight high-grade serous and five clear cell ovarian cancer genomes and we report here the identification of > 600 somatic rearrangements. Recurrent rearrangements of the transcriptional regulator gene, TSHZ3, were found in three of eight serous cases. Comparison to breast, pancreatic and prostate cancer genomes revealed that a subset of ovarian cancers share a marked tandem duplication phenotype with triple-negative breast cancers. The tandem duplication phenotype was not linked to BRCA1/2 mutation, suggesting that other common mechanisms or carcinogenic exposures are operative. High-grade serous cancers arising in women with germline BRCA1 or BRCA2 mutation showed a high frequency of small chromosomal deletions. These findings indicate that BRCA1/2 germline mutation may contribute to widespread structural change and that other undefined mechanism(s), which are potentially shared with triple-negative breast cancer, promote tandem chromosomal duplications that sculpt the ovarian cancer genome.

Exome sequencing identifies frequent mutation of the SWI/SNF complex gene PBRM1 in renal carcinoma.

The genetics of renal cancer is dominated by inactivation of the VHL tumour suppressor gene in clear cell carcinoma (ccRCC), the commonest histological subtype. A recent large-scale screen of ∼3,500 genes by PCR-based exon re-sequencing identified several new cancer genes in ccRCC including UTX (also known as KDM6A), JARID1C (also known as KDM5C) and SETD2 (ref. 2). These genes encode enzymes that demethylate (UTX, JARID1C) or methylate (SETD2) key lysine residues of histone H3. Modification of the methylation state of these lysine residues of histone H3 regulates chromatin structure and is implicated in transcriptional control. However, together these mutations are present in fewer than 15% of ccRCC, suggesting the existence of additional, currently unidentified cancer genes. Here, we have sequenced the protein coding exome in a series of primary ccRCC and report the identification of the SWI/SNF chromatin remodelling complex gene PBRM1 (ref. 4) as a second major ccRCC cancer gene, with truncating mutations in 41% (92/227) of cases. These data further elucidate the somatic genetic architecture of ccRCC and emphasize the marked contribution of aberrant chromatin biology.

Trichome density of main stem is tightly linked to PepMoV resistance in chili pepper (Capsicum annuum L.).

A relationship between pepper trichome and pepper mottle virus (PepMoV) resistance was examined. In an intraspecific F(2) mapping population from the cross between Capsicum annuum CM334 (trichome-bearing and PepMoV resistant) and Chilsungcho (glabrous and PepMoV susceptible), major QTLs for both traits were identified by composite interval mapping in linkage group (LG) 24 corresponding a telomere region on pepper chromosome 10. Ptel1 of putative trichome enhancing locus was a common major QTL for trichome density on the main stem and calyx. Ptel1 apart from HpmsE031 at a 1.03 cM interval was specifically associated to the trichome density on the main stem, whereas Ptel2 near m104 marker on LG2 was specific for the calyx trichome. Epistatic analysis indicated that Ptel1 engaged in controlling the trichome density by mutual interactions with the organ-specific QTLs. For PepMoV resistance, two QTLs (Pep1 and Pep2) were identified on the LG 24. Pep1 was located with Ptel1 in the R-gene cluster (RGC) for potyvirus resistance including Pvr4 with broad spectrum resistance to potyviruses. Pep1 flanking TG420 marker seemed to be the major factors determining correlation with PepMoV resistance. These results indicate that the level of trichome density on pepper main stem can be used as a morphological marker for Pvr4 in pepper breeding.