Genomes, variants, and massively parallel methods

Abstract

Introduction

It is easy to be carried away by the detectable peculiarities and to forget that much underlying variability is still hidden from view until some new technical device discloses the finer structure of chromosomes... Lionel Penrose, Chicago, IL. Third ISCN Consensus Conference, 1966

In 1966 it was recognized that the effort to characterize human cytogenetic variation was only the tip of the iceberg in terms of our understanding of genetic detail and that many more types of variation would be revealed with advancing technology. Since the time when DNA was discovered as the major molecule for genetic inheritance, there has been a need to understand how DNA variations affect growth, development, and disease. Even after over 50 years of advances in DNA technology, many types of DNA variation have yet to be identified, named, cataloged, and studied.

Human genome

The word genome signifies the collection of genes in an organism and is believed to have been coined by the German botanist Hans Winkler in the 1920s. The human genome encompasses all of the information needed for growth, development, and heredity. This information is copied in the nucleus of every cell in the body.

The determination of 46 chromosomes in humans occurred in 1956. The following years were marked by increased activity in human cytogenetics, but it soon became apparent that there was no coordination of how findings were named or classified. Beginning in 1960, a consensus meeting of laboratories (Denver Conference) established basic guidelines for naming large chromosomal variations. The findings from multiple subsequent consensus conferences were unified in a single document, “An International System for Human Cytogenetic Nomenclature (1978).”

Some of the basic concepts of chromosomes are described in ISCN 1978: autosomes are numbered from 1 to 22 in descending order of length. The sex chromosomes are named X and Y. The symbols p and q designated the short (p for “petit,” meaning small in French) and long arms of the chromosomes, respectively. A chromosome band is the part of the chromosome that is clearly distinguishable from adjacent segments which are darker or lighter in appearance. G-bands are the bands resulting from Giemsa dye staining. In addition to describing the normal state of chromosomal features, ISCN 1978 considered the naming of chromosomal rearrangements such as inversions, deletions, and translocations.

In more recent times, the ISCN 2013 version introduced new features such as the term “hg” for “human genome build or assembly,” and a chapter titled “Microarrays,” which is devoted to naming changes identified by oligonucleotide microarrays. Of note, there is a separate consortium focused on microarrays knows as ISCA (The International Standard for Cytogenomic Arrays). ISCA is focused on microarray test quality improvement by projects such as variant databases linked to clinical data.

Throughout the 1990s, there was an international effort to sequence the human genome. The first draft was released in 2001, ^, followed by a more complete version in 2004. The 2004 version contains 2.85 billion nucleotides (bases) and was considered 99% complete for euchromatic DNA. The overall size of the genome, including both euchromatic and heterochromatic sequence (tightly compact DNA found at centromeres and telomeres), was estimated to be 3.08 billion nucleotides. Thus the total overall genome was only 92.5% sequenced when it was first declared “essentially” complete. Within the 2.85 billion nucleotides of euchromatic DNA there were 19,438 known genes and an additional 2188 predicted genes. The total number of nucleotides encoding protein was approximately 34 million (1.2%) of the genome. This portion of the genome encoding proteins is also known as the exome . Genomic terms and definitions used in this chapter are given in Box 65.1 .

The 2004 genome contained 341 gaps in heterochromatic regions. These regions contain DNA that is difficult to sequence (e.g., repetitive elements, GC-rich sequence) or where no clone/template could be made. Commonly used DNA sequencing technologies require a scaffold on which sequence fragments are pieced together. The first human reference sequences were assembled by the University of California at Santa Cruz (UCSC) and were numbered starting with “hg1” in May of 2000. The National Center for Biotechnology Information (NCBI) produced their own genome builds starting in December 2001 as NCBI build 28 (equivalent to hg10 from UCSC) as the genome was further refined. This led to the publicly available 2004 version of the human genome known as NCBI35/hg17. This template or reference sequence has subsequently undergone continuous improvement under the international Genome Reference Consortium (GRC), producing GRCh37/hg19. In the future, only one designation will be given, such as the currently released GRCh38.

Since the 2004 genome publication, there have been continuing efforts to create “Platinum Genomes” that address the missing information (gaps) and improve the quality of data. Prior gaps have been sequenced by utilizing DNA from a haploid cell line. Long-read sequencing technologies can create de novo assemblies that do not require the use of reference genomes. Hybrid sequencing methods are emerging that combine the advantages of short-read sequencing for single nucleotide base accuracy with the advantages of long-read sequencing to further decrease gaps in human genome data and reveal new mechanisms of human variation. For example, the use of long-read sequencing technology has provided the first assembly of the highly repetitive centromere region of the Y-chromosome, and all gaps in the entire human X chromosome have been removed. Complete “telomere-to-telomere” genome sequencing of the malaria parasite, Plasmodium falciparum , has been accomplished. Eventually, it is likely that all human chromosomes will be sequenced from telomere-to-telomere, including the repetitive centromeric regions.

Each human cell contains two copies of the 3.08-billion genome divided into 46 chromosomes. Table 65.1 summarizes statistics for the human genome and the types of variations that are important in clinical diagnostics. Three quarters of human DNA is intergenic or between genes. More than 60% of this intergenic sequence consists of “parasitic” DNA regions of mostly defective transposable elements 100 to 11,000 bases in length. Between 2 and 3 million of these “retrotransposons” are present in each copy of the genome. They contribute to genetic recombination and chromosome structure and provide an evolutionary record of sequence variation and selection.

Segmental duplications constitute 5.3% of the human genome. They are over 1 kilobase (a thousand bases, or kb) in length, have a sequence identity of at least 90% and are not transposable. Segmental duplications are common in the human genome and are prone to deletion or rearrangements, often with medical consequences. Intergenic DNA also carries most of the simple sequence repeats (SSRs) present in the genome. A subset of SSRs, the short tandem repeats (STRs) have repeat units of 1 to several bases that may be repeated up to thousands of times. STRs have played a large role in genetic linkage studies and in forensic and medical identity testing. They are formed by slippage during replication and are highly polymorphic between individuals. The most common STRs are dinucleotide repeats, such as ACACAC and ATAT. On average, one STR occurs every 2000 bases.

Approximately 2% of DNA is required to maintain the structure of chromosomes and is located at chromosome centers (centromeres) and ends (telomeres) and makes up heterochromatic DNA. Centromeric DNA includes many tandem copies of nearly identical 171 base pair (bp) repeats encompassing 0.24 to 5.0 Mb per chromosome. Each chromosome end is capped with several kb of the telomeric 6 base repeat TTAGGG. Although intergenic DNA does not code for protein and was originally considered “junk,” much of this DNA is transcribed to RNA, producing a complex “transcriptome” network of RNA control elements whose function and mechanics are active areas of investigation.

There are about 20,000 genes that code for about 200,000 transcripts. Alternative splicing of exons produces many more transcripts than there are genes. The average gene covers 27,000 bases, but only about 1300 of these bases code for amino acids. The primary RNA transcript is processed by splicing to retain exons that are interspersed throughout the gene and have a higher GC content than noncoding regions. On average, 95% of a gene is excised as introns, retaining a mean of 10.4 exons, of which on average 9.1 are translated into proteins. Exons make up only 1.9% of the total genome, with 1.2% of the genome coding for proteins. Some important genes are present in many copies, so that overall protein expression is not affected if a chance variation occurs in one copy. If extra copies of genes lose their function, they are known as pseudogenes. At least as many pseudogenes as functional genes are present in the human genome. It is important to distinguish pseudogenes from functional genes because variants in pseudogenes are seldom of clinical importance, and they often complicate DNA diagnostic assays.

Nonhuman genomes

Before the human genome was completed, other genomes of smaller size were sequenced, enabling advancements in technology and logistical organization to sequence the human genome. ^, The genomes of different species vary in size and the complexity can be surprising. One of the largest known genomes is the white spruce tree ( Picea glauca ) at 26.9 billion bases. On the opposite end of the spectrum is Porcine circovirus-1, a single-stranded DNA virus with a genome that is less than 2000 bases. There is overlap in the genome size of eukaryotes (animals, plants, fungi), viruses, and bacteria ( Table 65.2 and Fig. 65.1 ).

DNA that codes for RNA but not protein

Even though 99% of the human genome does not code for protein, most of it is transcribed into noncoding RNA. At least 93% of the genome is transcribed, producing more than 10 times the amount of RNA that is produced from the coding segments of genes. Both strands of DNA may be transcribed, and long noncoding transcripts may overlap coding regions, producing a complex transcriptome of functional RNA molecules that may variably regulate transcription of coding regions, RNA processing, mRNA stability, translation, protein stability, and secretion. In addition to long noncoding RNA, ribosomal RNA, and transfer RNA, specific classes of noncoding RNAs include small nuclear RNAs critical for splicing, small nucleolar RNAs that modify rRNA, telomerase RNAs for maintenance of telomeres, small interfering RNAs, and microRNAs (miRNAs) that regulate gene expression. In a recent review on RNA, 54 different categories were identified. Some of the more important types of RNA are listed in Table 65.3 .

MicroRNAs (or miRNAs) are particularly interesting as potential markers for disease. For example, concentrations of specific, circulating miRNAs correlate with many different types of cancer. MicroRNAs are noncoding but functional single-stranded RNAs that are 21 to 22 bases long and are expressed in a tissue-specific manner. They are initially transcribed as longer precursors that undergo two rounds of truncations as they are transported from the nucleus to the cytoplasm in the cell. The mature miRNA is then integrated into a protein complex called the RNA-induced silencing complex , which regulates translation of mRNA. MicroRNAs hybridize to a 6 to 8 base sequence in the 3′ untranslated region of target mRNAs and inhibit mRNA expression either by mRNA degradation if the bases are perfectly complementary, or by blocking of translation if they are imperfectly complementary. Currently for humans, there are 1917 precursor miRNA and 2654 mature miRNAs cataloged in miRBase. Despite the promise of miRNAs as tumor markers, the literature is often contradictory and inconsistent with few accepted conclusions.

Variation in the human genome

If the DNA of any two individuals is compared, on average one difference is noted every 1250 bases (i.e. , approximately 99.9% of the sequence is identical between randomly chosen copies of the genome). However, copy-number variants involve a greater amount of the genome, with 0.5% of the genome differing on average between two individuals when copy-number variants greater than 50 kb are considered. Between individuals, at least five times as many bases are affected by copy number changes as by small sequence differences.

Most human genetic material is present in two copies, with the exception of the unpaired sex chromosome in males and mitochondrial DNA. The presence of only single gene copies on the X and Y chromosome in males leads to well-known sex-linked disorders. In contrast, the 16,500-bp mitochondrial genome is present in multiple copies per cell, constituting about 0.3% of human DNA, depending on the tissue source. Allele fractions may vary over a wide range when all mitochondria in a cell are considered. That is, sequence variations in mitochondrial DNA are heteroplasmic, meaning that the ratio of the wild-type allele to a variant allele can vary almost continuously, sometimes resulting in a wide range of symptoms even when only one sequence variant is involved.

Large-scale human genome sequencing projects have cast a wide net across many diverse populations. These projects have provided a wealth of knowledge of the genetic diversity that exists in humans. An alternative approach to human genetic diversity is to examine more homogenous populations. Several studies have examined the genetics of a large number of individuals from Iceland. A whole genome sequencing study of 2636 Icelanders observed 20 million single nucleotide variants (SNVs) and 1.5 million insertions/deletions. The data from this whole genome sequencing study were combined with a previous data set of 104,220 Icelanders who had been SNV typed at 676,913 locations. By applying whole genome sequencing data from only a small subset of individuals, the full genetics could be inferred for a larger set of over 100,000 individuals who had only had SNV typing.

Another interesting result of the Icelandic whole genome study was the identification of 6795 loss of function single nucleotide variants, insertions, or deletions in 4924 genes. Loss of function changes (homozygous or compound heterozygous) were found in 7.7% of the individuals sequenced. In essence, this study identified a surprisingly high percentage of individuals with “knocked-out” or functionally silenced genes.

Any sequence change (compared to a reference sequence) is called a sequence variant or variation. Many variations do not affect human health and are benign or silent. For example, most (1) copy-number variations, (2) SNVs, and (3) STRs found between genes are seldom associated with disease.