Lorenzo Tattini

EXCAVATOR: detecting copy number variants from whole-exome sequencing data.

AbstractWe developed a novel software tool, EXCAVATOR, for the detection of copy number variants (CNVs) from whole-exome sequencing data. EXCAVATOR combines a three-step normalization procedure with a novel heterogeneous hidden Markov model algorithm and a calling method that classifies genomic regions into five copy number states. We validate EXCAVATOR on three datasets and compare the results with three other methods. These analyses show that EXCAVATOR outperforms the other methods and is therefore a valuable tool for the investigation of CNVs in largescale projects, as well as in clinical research and diagnostics. EXCAVATOR is freely available at http://sourceforge.net/projects/excavatortool/.

Detection of Genomic Structural Variants from Next-Generation Sequencing Data.

Structural variants are genomic rearrangements larger than 50 bp accounting for around 1% of the variation among human genomes. They impact on phenotypic diversity and play a role in various diseases including neurological/neurocognitive disorders and cancer development and progression. Dissecting structural variants from next-generation sequencing data presents several challenges and a number of approaches have been proposed in the literature. In this mini review, we describe and summarize the latest tools – and their underlying algorithms – designed for the analysis of whole-genome sequencing, whole-exome sequencing, custom captures, and amplicon sequencing data, pointing out the major advantages/drawbacks. We also report a summary of the most recent applications of third-generation sequencing platforms. This assessment provides a guided indication – with particular emphasis on human genetics and copy number variants – for researchers involved in the investigation of these genomic events.

Read count approach for DNA copy number variants detection

MOTIVATION The advent of high-throughput sequencing technologies is revolutionizing our ability in discovering and genotyping DNA copy number variants (CNVs). Read count-based approaches are able to detect CNV regions with an unprecedented resolution. Although this computational strategy has been recently introduced in literature, much work has been already done for the preparation, normalization and analysis of this kind of data. RESULTS Here we face the many aspects that cover the detection of CNVs by using read count approach. We first study the characteristics and systematic biases of read count distributions, focusing on the normalization methods designed for removing these biases. Subsequently, we compare the algorithms designed to detect the boundaries of CNVs and we investigate the ability of read count data to predict the exact number of DNA copy. Finally, we review the tools publicly available for analysing read count data. To better understand the state of the art of read count approaches, we compare the performance of the three most widely used sequencing technologies (Illumina Genome Analyzer, Roche 454 and Life Technologies SOLiD) in all the analyses that we perform.

H3M2: detection of runs of homozygosity from whole-exome sequencing data

MOTIVATION Runs of homozygosity (ROH) are sizable chromosomal stretches of homozygous genotypes, ranging in length from tens of kilobases to megabases. ROHs can be relevant for population and medical genetics, playing a role in predisposition to both rare and common disorders. ROHs are commonly detected by single nucleotide polymorphism (SNP) microarrays, but attempts have been made to use whole-exome sequencing (WES) data. Currently available methods developed for the analysis of uniformly spaced SNP-array maps do not fit easily to the analysis of the sparse and non-uniform distribution of the WES target design. RESULTS To meet the need of an approach specifically tailored to WES data, we developed [Formula: see text], an original algorithm based on heterogeneous hidden Markov model that incorporates inter-marker distances to detect ROH from WES data. We evaluated the performance of [Formula: see text] to correctly identify ROHs on synthetic chromosomes and examined its accuracy in detecting ROHs of different length (short, medium and long) from real 1000 genomes project data. [Formula: see text] turned out to be more accurate than GERMLINE and PLINK, two state-of-the-art algorithms, especially in the detection of short and medium ROHs. AVAILABILITY AND IMPLEMENTATION [Formula: see text] is a collection of bash, R and Fortran scripts and codes and is freely available at https://sourceforge.net/projects/h3m2/. CONTACT albertomagi@gmail.com SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

Characterisation of mutations of the phosphoinositide-3-kinase regulatory subunit, PIK3R2, in perisylvian polymicrogyria: a next-generation sequencing study

SUMMARY Background Bilateral perisylvian polymicrogyria (BPP), the most common form of regional polymicrogyria, causes the congenital bilateral perisylvian syndrome, featuring oromotor dysfunction, cognitive impairment and epilepsy. BPP is etiologically heterogeneous, but only a few genetic causes have been reported. The aim of this study was to identify additional genetic etiologies of BPP and delineate their frequency in this patient population. Methods We performed child-parent (trio)-based whole exome sequencing (WES) on eight children with BPP. Following the identification of mosaic PIK3R2 mutations in two of these eight children, we performed targeted screening of PIK3R2 in a cohort of 118 children with BPP who were ascertained from 1980 until 2015 using two methods. First, we performed targeted sequencing of the entire PIK3R2 gene by single molecule molecular inversion probes (smMIPs) on 38 patients with BPP with normal-large head size. Second, we performed amplicon sequencing of the recurrent PIK3R2 mutation (p.Gly373Arg) on 80 children with various types of polymicrogyria including BPP. One additional patient underwent clinical WES independently, and was included in this study given the phenotypic similarity to our cohort. All patients included in this study were children (< 18 years of age) with polymicrogyria enrolled in our research program. Findings Using WES, we identified a mosaic mutation (p.Gly373Arg) in the regulatory subunit of the PI3K-AKT-MTOR pathway, PIK3R2, in two children with BPP. Of the 38 patients with BPP and normal-large head size who underwent targeted next generation sequencing by smMIPs, we identified constitutional and mosaic PIK3R2 mutations in 17 additional children. In parallel, one patient was found to have the recurrent PIK3R2 mutation by clinical WES. Seven patients had BPP alone, and 13 had BPP in association with features of the megalencephaly-polymicrogyria-polydactyly-hydrocephalus syndrome (MPPH). Nineteen patients had the same mutation (Gly373Arg), and one had a nearby missense mutation (p.Lys376Glu). Across the entire cohort, mutations were constitutional in 12 and mosaic in eight patients. Among mosaic patients, we observed substantial variation in alternate (mutant) allele levels ranging from 2·5% (10/377) to 36·7% (39/106) of reads, equivalent to 5–73·4% of cells analyzed. Levels of mosaicism varied from undetectable to 17·1% (37/216) of reads in blood-derived compared to 29·4% (2030/6889) to 43·3% (275/634) in saliva-derived DNA. Interpretation Constitutional and mosaic mutations in the PIK3R2 gene are associated with a spectrum of developmental brain disorders ranging from BPP with a normal head size to the megalencephaly-polymicrogyria-polydactyly-hydrocephalus syndrome. The phenotypic variability and low-level mosaicism challenging conventional molecular methods have important implications for genetic testing and counseling.

Enhanced copy number variants detection from whole-exome sequencing data using EXCAVATOR2.

Copy Number Variants (CNVs) are structural rearrangements contributing to phenotypic variation that have been proved to be associated with many disease states. Over the last years, the identification of CNVs from whole-exome sequencing (WES) data has become a common practice for research and clinical purpose and, consequently, the demand for more and more efficient and accurate methods has increased. In this paper, we demonstrate that more than 30% of WES data map outside the targeted regions and that these reads, usually discarded, can be exploited to enhance the identification of CNVs from WES experiments. Here, we present EXCAVATOR2, the first read count based tool that exploits all the reads produced by WES experiments to detect CNVs with a genome-wide resolution. To evaluate the performance of our novel tool we use it for analysing two WES data sets, a population data set sequenced by the 1000 Genomes Project and a tumor data set made of bladder cancer samples. The results obtained from these analyses demonstrate that EXCAVATOR2 outperforms other four state-of-the-art methods and that our combined approach enlarge the spectrum of detectable CNVs from WES data with an unprecedented resolution. EXCAVATOR2 is freely available at http://sourceforge.net/projects/excavator2tool/.

Characterization of MinION nanopore data for resequencing analyses

The Oxford Nanopore Technologies MinION is a new device, based on nanopore sequencing that is able to generate reads of tens of kilobases in length with faster sequencing time with respect to other platforms. To evaluate the capability of nanopore data to be exploited for resequencing analyses we used the largest MinION data set to date and we compared with Illumina and Pacific Biosciences technologies. By using five different mapping approaches we estimated that the global sequencing error rate of MinION reads, mainly caused by inserted and deleted bases, is around 11%. The study of error distribution showed that substituted, inserted and deleted bases are not randomly distributed along the reads, but mainly occur in specific nucleotide patterns, generating a significant number of genomic loci that can be misclassified as false-positive variants. With 40× sequencing coverage, MinION data can produce at best around one false substitution and insertion every 10-50 kb, and one false deletion every 1000 bp, making use of this technology still challenging for small-sized variant discovery. We also analyzed depth of coverage distribution and we demonstrated that nanopore sequencing is a uniform process that generates sequences randomly and independently without classical sources of bias such as GC-content and mappability. Owing to these properties, the MinION data can be readily used to detect genomic regions involved in copy number variants with high accuracy, outperforming other state-of-the-art sequencing methods in terms of both sensitivity and specificity.

Characterization and identification of hidden rare variants in the human genome

BackgroundBy examining the genotype calls generated by the 1000 Genomes Project we discovered that the human reference genome GRCh37 contains almost 20,000 loci in which the reference allele has never been observed in healthy individuals and around 70,000 loci in which it has been observed only in the heterozygous state.ResultsWe show that a large fraction of this rare reference allele (RRA) loci belongs to coding, functional and regulatory elements of the genome and could be linked to rare Mendelian disorders as well as cancer. We also demonstrate that classical germline and somatic variant calling tools are not capable to recognize the rare allele when present in these loci. To overcome such limitations, we developed a novel tool, named RAREVATOR, that is able to identify and call the rare allele in these genomic positions. By using a small cancer dataset we compared our tool with two state-of-the-art callers and we found that RAREVATOR identified more than 1,500 germline and 22 somatic RRA variants missed by the two methods and which belong to significantly mutated pathways.ConclusionsThese results show that, to date, the investigation of around 100,000 loci of the human genome has been missed by re-sequencing experiments based on the GRCh37 assembly and that our tool can fill the gap left by other methods. Moreover, the investigation of the latest version of the human reference genome, GRCh38, showed that although the GRC corrected almost all insertions and a small part of SNVs and deletions, a large number of functionally relevant RRAs still remain unchanged. For this reason, also future resequencing experiments, based on GRCh38, will benefit from RAREVATOR analysis results. RAREVATOR is freely available at http://sourceforge.net/projects/rarevator.

WNP: A Novel Algorithm for Gene Products Annotation from Weighted Functional Networks

Predicting the biological function of all the genes of an organism is one of the fundamental goals of computational system biology. In the last decade, high-throughput experimental methods for studying the functional interactions between gene products (GPs) have been combined with computational approaches based on Bayesian networks for data integration. The result of these computational approaches is an interaction network with weighted links representing connectivity likelihood between two functionally related GPs. The weighted network generated by these computational approaches can be used to predict annotations for functionally uncharacterized GPs. Here we introduce Weighted Network Predictor (WNP), a novel algorithm for function prediction of biologically uncharacterized GPs. Tests conducted on simulated data show that WNP outperforms other 5 state-of-the-art methods in terms of both specificity and sensitivity and that it is able to better exploit and propagate the functional and topological information of the network. We apply our method to Saccharomyces cerevisiae yeast and Arabidopsis thaliana networks and we predict Gene Ontology function for about 500 and 10000 uncharacterized GPs respectively.

Exome sequencing of two Italian pedigrees with non-isolated Chiari malformation type I reveals candidate genes for cranio-facial development

Chiari malformation type I (CMI) is a congenital abnormality of the cranio-cerebral junction with an estimated incidence of 1 in 1280. CMI is characterized by underdevelopment of the occipital bone and posterior fossa (PF) and consequent cerebellar tonsil herniation. The presence for a genetic basis to CMI is supported by many lines of evidence. The cellular and molecular mechanisms leading to CM1 are poorly understood. The occipital bone formation is dependent on complex interactions between genes and molecules with pathologies resulting from disruption of this delicate process. Whole-exome sequencing of affected and not affected individuals from two Italian families with non-isolated CMI was undertaken. Single-nucleotide and short insertion–deletion variants were prioritized using KGGSeq knowledge-based platform. We identified three heterozygous missense variants: DKK1 c.121G>A (p.(A41T)) in the first family, and the LRP4 c.2552C>G (p.(T851R)) and BMP1 c.941G>A (p.(R314H)) in the second family. The variants were located at highly conserved residues, segregated with the disease, but they were not observed in 100 unaffected in-house controls. DKK1 encodes for a potent soluble WNT inhibitor that binds to LRP5 and LRP6, and is itself regulated by bone morphogenetic proteins (BMPs). DKK1 is required for embryonic head development and patterning. LRP4 is a novel osteoblast expressed receptor for DKK1 and a WNT and BMP 4 pathways integrator. Screening of DKK1 in a cohort of 65 CMI sporadic patients identified another missense variant, the c.359G>T (p.(R120L)), in two unrelated patients. These findings implicated the WNT signaling in the correct development of the cranial mesenchyme originating the PF.