A term baby is born with a cleft palate, low-set ears and clenched hands with overlapping fingers. The paediatric team is asked a deceptively simple question: does this baby need a genetic test, and if so, which one? Getting that answer right — and in the right order — increasingly decides whether a family leaves the unit with a diagnosis or with uncertainty.
Why this matters
Congenital anomalies and genetic disorders are not a rare curiosity at the edge of neonatal practice; they sit close to its centre. Significant structural malformations affect roughly 3% to 6% of all live births, which works out at about 8 million children born each year with a severe birth defect, and an estimated 5% to 10% of newborns carry a suspected disease-causing chromosomal disorder.
As we get better at keeping preterm and infected babies alive, congenital anomalies are steadily rising up the list of what actually causes neonatal and infant death — in some datasets up to 50% of infant deaths ultimately trace back to an underlying genetic disorder. The diagnostic toolkit has expanded just as fast, and choosing well among karyotype, microarray, gene panels and genome-wide sequencing is now a core clinical skill for everyone around the cot, not just the geneticist.
The scale of the problem
The World Health Organization estimates that congenital anomalies contribute to around 303,000 neonatal deaths worldwide each year. In the United States, chromosomal and congenital anomalies account for about 20% of all infant deaths, and they weigh even more heavily inside the neonatal intensive care unit (NICU), where affected infants are hospitalised more often, for longer, and at greater cost than babies without an underlying genetic defect.
The burden is not spread evenly. It falls disproportionately on low- and middle-income countries, driven by limited access to prenatal screening and advanced care, widespread nutritional deficiencies such as inadequate folic acid, and higher exposure to environmental teratogens and infection. India illustrates the point: with a large population and a deep, diverse genetic pool, roughly 1 in 33 infants is born with a congenital anomaly — about 1.7 million affected births a year, led by neural tube defects, congenital heart defects and chromosomal abnormalities.
One quiet driver of the “rising” prevalence is worth naming: as next-generation sequencing becomes cheaper and more capable, we are simply finding more of what was always there. Some of the apparent increase in genetic-variant-linked disease is better detection, not a true rise in incidence.
Genome versus exome — the distinction that shapes the test menu
Before ordering anything, it helps to be clear about what each test actually reads. The genome is the entirety of a person’s DNA — around 3 billion base pairs, encompassing both the protein-coding regions (exons) and the non-coding regions in between (introns). The exome is a small, specific subset: the exons alone, the stretches of DNA actively translated into protein. Despite carrying most known disease-causing mutations, the exome accounts for only about 1% to 2% of the genome — roughly 50 million base pairs.
That single fact explains much of what follows. Whole exome sequencing (WES) deliberately reads only that coding 1% to 2%, so it captures most known protein-changing mutations but misses variation buried in the deep intronic regions. Whole genome sequencing (WGS) reads the DNA end to end, capturing both the exome and the remaining 98% to 99% of non-coding sequence — territory we increasingly recognise as important for regulating how and when genes are switched on.
The detailed assessment comes first
The temptation with a powerful test menu is to reach for it early. The source is clear that the first step is not a test at all, but a structured clinical assessment that tells you whether a genetic cause is even in play and, if so, which test will have the highest yield.
History gathering sets the frame. Note parental age, ethnicity and any consanguinity; take a careful obstetric history including prior terminations, stillbirths and previously affected children. Review first-trimester exposures deliberately — over-the-counter medicines, anticonvulsants, acne treatments, illicit drugs, radiation, any febrile illness with a rash, and the mother’s vaccination status for infections such as rubella and varicella. Where possible, construct a pedigree spanning up to four generations. If there is an affected relative — the index case — document age of onset, disease progression and cause of death, and collect clinical photographs (front and back of the body, facial profile, hands and feet) and videos of any abnormal movement, gait or behaviour to help with later pattern recognition.
Counselling should be multidisciplinary and staged. When an anomaly is suspected antenatally, an obstetrician, fetal medicine specialist, neonatologist, medical geneticist and relevant paediatric subspecialists should begin preparing the family early — for the range of outcomes, for the interventions that may be needed at delivery, and, where a condition is incompatible with survival, for comfort care. The team should lay out the prenatal options (amniocentesis, chorionic villus sampling) and postnatal testing routes honestly, including their limitations and the real possibility of an inconclusive result.
The postnatal examination is systematic and head-to-toe. Skull size and shape, hairline and scalp defects; eyes, ears, nasal bridge and oral structures; neck webbing and cystic hygromas; chest wall, organomegaly, herniae and spinal markers such as sacral dimples or hair tufts; limbs for shortening, polydactyly, syndactyly and abnormal palmar creases; skin for café-au-lait or ash-leaf macules; genitalia and anus for ambiguity, hypospadias, cryptorchidism and patency. Because sensory and endocrine deficits track closely with genetic conditions, a detailed eye examination, audiologic testing and thyroid function tests belong in the workup of any dysmorphic neonate, alongside targeted imaging and biochemistry driven by the presentation.
Building a four-generation pedigree — and why consanguinity matters
The pedigree is only as useful as the index case documentation behind it: age at symptom onset, growth or developmental delay, the pattern of disease progression, and, if the person has died, exactly how. Clinical photographs and — where speech, behaviour, gait or limb movements are abnormal — video recordings help the team recognise phenotypic patterns that are hard to hold in memory.
Consanguinity deserves explicit attention. Reproduction between close relatives, such as first cousins, sharply raises the chance that both parents are silent carriers of the same recessive mutation, and so raises the risk of recessive disease in the child. Tracing these relationships across an extended pedigree can reveal an autosomal recessive pattern that would otherwise look sporadic. The clinical significance is not new: alkaptonuria was linked to heredity over a century ago precisely because it clustered among the children of first cousins.
When the cause is the environment, not the genes
Not every anomaly is genetic, and telling the two apart changes both testing and counselling. Environmental exposures and teratogens are external, non-genetic factors that interfere with normal development. They tend to produce disruptions — for example, missing limbs or digits from amniotic bands or vascular interruption — or deformations, such as clubfoot or arthrogryposis caused by the mechanical constraint of oligohydramnios.
The categories worth investigating are well defined. Medications and drugs in the first trimester, including anticonvulsants, isotretinoin, thalidomide, warfarin and antiretrovirals. Maternal health and nutrition — uncontrolled diabetes (especially early), obesity, thyroid disease and hypercholesterolaemia, alongside deficiencies such as low folic acid (strongly tied to neural tube defects) and, in animal models, vitamin A deficiency linked to congenital diaphragmatic hernia. Intrauterine infection, chiefly the TORCH group — rubella, cytomegalovirus, toxoplasmosis and varicella — which is why vaccination history and any first-trimester febrile rash matter. And pollutants, toxins and radiation: maternal alcohol as a primary teratogen; dioxins, PCBs and agricultural pesticides linked to congenital heart disease; specific agents such as phenmetrazine, quinine and nitrofen associated with diaphragmatic hernia; and prolonged radiation exposure.
The clinical payoff is direct. If an anomaly is confidently attributed to a teratogen or environmental exposure, genetic testing is usually not warranted. And the counselling shifts: unlike an inherited mutation, the recurrence risk for an environmentally induced anomaly is not fixed — it depends entirely on whether the same exposure recurs in a future pregnancy.
Morphological characterisation — naming the defect correctly
Categorising a defect by its developmental mechanism is not academic tidiness; it steers the decision to test and the choice of test. The source sets out a working vocabulary.
Normal variations are minor departures from the reference population, often harmless — low-set ears, hypertelorism — but occasionally flagging a major anomaly. Malformations are congenital, non-progressive abnormalities arising when the primary programme of morphogenesis is interrupted; they split into major malformations with serious medical, surgical or developmental consequences (central nervous system anomalies, congenital heart disease) and minor ones that rarely cause problems alone (an extra digit, a single transpalmar crease). The clinical value of noting both is that a particular combination of major and minor malformations can point to a single underlying syndrome.
Deformations are previously normal structures distorted by mechanical force, such as clubfoot or arthrogryposis from oligohydramnios. Disruptions occur when normal development is interrupted by an external event — vascular interruption, amniotic bands, teratogens — producing, for example, missing limbs. Dysplasias stem from abnormal proliferation and organisation of cells within a tissue, as in skeletal or ectodermal dysplasias.
Above the level of the single defect, the team should look for recognisable groupings. A syndrome is a set of major and minor anomalies tied to one underlying cause — Edward syndrome (trisomy 18), Cornelia de Lange syndrome. A sequence is a cascade of secondary anomalies flowing from one initial problem, classically the Pierre Robin sequence, where an underdeveloped jaw displaces the tongue and produces a cleft palate. An association is a cluster of anomalies that co-occur more often than chance but without a single unifying cause yet identified — the VACTERL association and the OEIS complex are the standard examples.
Pattern recognition — from phenotype to the right test
With the examination and morphological characterisation complete, the third step is to decide whether the findings are isolated or whether multiple anomalies cluster into a recognisable pattern consistent with a chromosomal defect, a microdeletion or duplication, or a single-gene disorder. This is the step that most directly determines which test to order.
Certain clusters point strongly toward common aneuploidies. Trisomy 21 (Down syndrome) often shows increased nuchal translucency, an absent or hypoplastic nasal bone, atrioventricular septal defects and duodenal atresia (the “double bubble”). Trisomy 18 (Edward syndrome) brings a strawberry-shaped skull, choroid plexus cysts, clenched hands with overlapping index fingers and rocker-bottom feet. Trisomy 13 (Patau syndrome) is recognised by severe midline facial abnormalities, holoprosencephaly and postaxial polydactyly. Triploidy carries a pattern of CNS anomalies, severe facial defects and complex cardiac lesions such as tetralogy of Fallot.
Because the number of genes tied to rare disease runs into the thousands, no clinician can carry every pattern in their head. Dysmorphology databases exist for exactly this: the Winter-Baraitser Dysmorphology Database, Online Mendelian Inheritance in Man (OMIM), Pictures of Standard Syndromes and Undiagnosed Malformations, and Phenomizer. OMIM is a particularly central hub — a daily-updated, searchable catalogue of more than 16,000 described genes, including over 7,000 with known phenotypic associations and over 4,500 with specific phenotype-causing mutations, and it links out to ClinVar, the public archive of evidence linking human variants to phenotypes. To put its reach in perspective, WGS can now diagnose most of the more than 6,000 conditions catalogued in OMIM.
The recognised pattern then dictates the algorithm. A picture suggesting a common trisomy, or a baby with ambiguous genitalia, usually warrants a targeted karyotype. Multiple unexplained anomalies or developmental delay without a clear single cause point to chromosomal microarray (CMA) as first-line. A pattern suggesting a specific monogenic disorder, a skeletal dysplasia or a neuromuscular condition justifies bypassing the karyotype and going straight to next-generation sequencing (WES or WGS).
The testing toolkit, test by test
Karyotype
The traditional first-line test reports the number and gross structure of the chromosomes, detecting aneuploidies, large translocations and large insertions or deletions. It earns its place for suspected common trisomies (13, 18, 21), disorders of sex development such as ambiguous genitalia, and — importantly — couples with two or more first-trimester miscarriages, where it can reveal a balanced translocation that a microarray cannot. It is also recommended for specific major malformations (tetralogy of Fallot, atrioventricular septal defect, diaphragmatic hernia, omphalocele, Dandy-Walker malformation, bilateral clubfoot) and for multiple anomalies with early-onset growth restriction.
Its limitation is resolution. A karyotype only sees missing or extra segments of 4 to 5 million base pairs or larger, so it cannot detect submicroscopic microdeletions and microduplications — which is why its overall diagnostic yield is only around 5%, and why CMA is preferred when a submicroscopic abnormality is suspected. Turnaround is typically 14 to 21 days. Equally, it is not required for isolated anomalies where a chromosomal cause is unlikely, such as isolated hypoplastic left heart syndrome, isolated neural tube defects, unilateral isolated clubfoot or amniotic band syndrome.
Targeted gene panels
Panels use NGS to sequence a predefined set of genes known to cause a particular disease, usually focusing on the coding exons and nearby regulatory regions. They suit a neonate with a specific, recognisable phenotype — dedicated panels exist for neonatal respiratory distress, seizures or hypotonia — and within their gene set they detect single nucleotide variants, small indels and copy number changes well.
Their weakness is that they are not agnostic. Scope varies dramatically between laboratories; a neonatal respiratory panel might cover as few as 5 core genes (ABCA3, FOXF1, NKX2-1, SFTPB, SFTPC) or screen over 100. Because the panel only looks where the clinician tells it to, it depends on correctly guessing the culprit genes from the presentation. That is a real problem in the NICU, where babies often present with vague signs or have not yet manifested the full classic phenotype — if the symptoms do not align with the preselected genes, the true diagnosis is missed.
Chromosomal microarray (CMA)
CMA scans the whole genome for copy number variants — small gains and losses of genetic material — and has become the preferred first-tier test for CNVs in complex neonatal cases: multiple malformations, unexplained growth restriction or failure to thrive, and unexplained intellectual disability with or without dysmorphism. Using oligonucleotide and single-nucleotide probes, it needs no cell culture (making it more robust than karyotyping), turns around in 10 to 14 days, and reaches far finer resolution — deletions as small as 50 to 100 kilobases and duplications up to 400 kilobases, against the karyotype’s 4 to 5 megabase floor.
That resolution lifts the diagnostic yield to roughly 12% to 20%, and up to 23.9% in specific populations such as pregnancies terminated for fetal malformation — though still below the pooled 38.6% to 41% of WGS. The blind spots matter: because CMA only detects missing or extra material, it cannot see balanced rearrangements such as balanced translocations or inversions, it cannot resolve CNVs smaller than 50 kb, and its genome-wide reach means it frequently turns up variants of uncertain significance (VUS), which often require parental testing to establish whether a variant was inherited or arose de novo.
Whole exome sequencing (WES)
WES targets the protein-coding 1% to 2% of the genome and reads it down to the individual variant — point mutations and single nucleotide variants. It comes into its own when a neonate has multiple congenital anomalies and karyotype or CMA have not delivered a diagnosis, and is especially high-yielding when the picture suggests a monogenic disorder such as an inborn error of metabolism, a skeletal dysplasia, a neuromuscular disorder, or multi-system involvement.
The numbers are compelling: in structurally abnormal fetuses with a normal CMA, WES adds a diagnostic yield of about 31%, and in critically ill infants and children meta-analyses put the pooled yield at 36% to 37.8%. Standard turnaround is 4 to 6 weeks, but rapid WES (rWES) can return results in roughly 11 days — matching rapid WGS. The trade-offs follow from its targeted nature: it misses deep intronic variants, struggles with large structural variants and CNVs whose boundaries fall outside single exons, may miss small indels, and — because of sequencing technique — offers less depth of coverage for SNVs even within the exons than WGS. It remains, nonetheless, a cost-effective option that still captures the great majority of known disease-causing protein mutations.
Whole genome sequencing (WGS)
WGS is the most comprehensive single test in clinical practice, reading the exons, introns and the mitochondrial genome in one go. In a single run it can detect SNVs, small indels, CNVs and aneuploidies, and it can diagnose most of the 6,000-plus OMIM conditions. Pooled diagnostic yield sits at 38.6% to 41%, comfortably ahead of standard-of-care testing such as microarray (around 7.8% to 20%). Recent guidance positions WGS (or WES) as a first- or second-line test for neonates and infants with congenital anomalies or developmental delay, and it is particularly valuable when the phenotype is complex or incomplete and no single disorder can be confidently targeted.
Its advantages over WES are concrete: it reads deep intronic regulatory variants WES misses, offers superior coverage of SNVs even within exons, and detects larger structural variants and CNVs spanning beyond a single exon. But it is not omniscient, and most of its blind spots stem from current short-read technology, which fragments DNA into 50 to 300 base-pair pieces before reading. Short-read WGS struggles to count identical repetitive sequences — it often misses the SMN1 deletion causing spinal muscular atrophy because of the near-identical SMN2 backup gene, and some laboratories cannot detect trinucleotide repeat expansions such as the one behind myotonic dystrophy. It cannot read methylation or packaging, so it misses imprinting disorders such as Prader-Willi and Beckwith-Wiedemann syndromes. Like CMA, it cannot easily see balanced translocations. And because its coverage depth across 3 billion base pairs is lower than a panel’s, it is less reliable for mosaicism and low-level mitochondrial heteroplasmy.
WGS also raises findings the family did not go looking for. Incidental findings turn up by chance — uncovering Klinefelter syndrome while testing for an isolated cardiac defect. Secondary findings are known, medically actionable adult-onset risks, such as a BRCA1 variant or a familial cardiovascular mutation. Because these carry long-term implications for the infant and potentially the parents, thorough genetic counselling before ordering WGS — with an explicit opt-in or opt-out — is essential.
Rapid and ultra-rapid WGS in the NICU
For a critically ill baby, time is the variable that changes outcomes. Rapid WGS (rWGS) aims to return actionable results in under two weeks; ultra-rapid WGS (urWGS) can deliver preliminary reports in days or even hours. In critically ill neonates, early studies reported diagnostic yields of 42% to 57%, and — more to the point — these diagnoses altered clinical management in 30% to 72% of patients and changed outcomes in 24% to 34%.
What that looks like at the bedside is specific. A rapid diagnosis can start a targeted therapy — fosdenopterin for molybdenum cofactor deficiency — or direct surgery such as tracheostomy or mandibular distraction. It can redirect goals of care: identifying a lethal variant such as a FOXF1 mutation causing alveolar capillary dysplasia may guide the team to withdraw ECMO or move to palliative care. Even a negative result carries weight; while it cannot exclude every genetic cause, it eliminates thousands of possibilities, eases parental anxiety and lets the team commit to a plan built around a likely non-genetic cause.
Several trials underpin this. In NSIGHT1, a 26-hour rWGS system diagnosed 41% of infants versus 24% with standard testing. NSIGHT2 offered ultra-rapid WGS to gravely ill infants and reached a 46% diagnostic rate with a median turnaround of 4.6 days. NICU-Seq, comparing a 15-day WGS protocol against standard care, found roughly twice the rate of management change by 60 days (21.1% versus 10.3%). GEMINI, comparing rWGS against targeted neonatal panels, generated preliminary reports within 72 hours for urgent cases and identified structural variants and changes in 8 genes that the panels missed entirely. Reflecting all this, the American College of Medical Genetics and Genomics (ACMG) recommends WES or WGS as a first- or second-line test for congenital anomalies or developmental delay, and, for critically ill NICU patients, that it be ordered rapidly with a target turnaround under two weeks.
The trio advantage — and its limits
For both WES and WGS, analysing the baby alongside both biological parents (a trio) improves accuracy by immediately showing whether a variant is de novo or inherited — central to judging pathogenicity and recurrence risk. But it is not without friction. One or both parents may be unavailable, in which case proband-only testing should still proceed, as some laboratories report near-equivalent yield. It is a misconception that trio analysis is a full workup of the parents; their data is used only to interpret the infant’s candidate variants and is not a substitute for formal parental testing. Because it directly compares parental and child DNA, it can surface sensitive findings such as non-paternity. And the marginal yield gain can be modest — NSIGHT2 found that trio analysis improved the diagnosis for only 1 of 147 infants over proband-only testing, though it did remove the need for secondary confirmatory sequencing.
Assessing recurrence risk
Once a diagnosis is in hand, counselling turns to the next pregnancy. Recurrence risk — the probability that the abnormality recurs — lets families weigh reproductive options such as prenatal testing or preimplantation genetic diagnosis, start preventive measures like folic acid, and prepare practically. It depends on the mechanism.
For single-gene (Mendelian) disorders, the numbers follow the inheritance pattern. Autosomal dominant: a 50% chance from an affected parent, but under 1% if the variant arose de novo and neither parent is affected. Autosomal recessive: with two carrier parents, 25% affected, 50% carrier, 25% unaffected non-carrier — with consanguinity raising the stakes. X-linked dominant: an affected mother passes it to 50% of children; an affected father to all daughters and no sons. X-linked recessive: a carrier mother has a 50% chance of an affected son and a 50% chance of a carrier daughter, while an affected father makes all daughters carriers and leaves sons unaffected.
For chromosomal abnormalities, the mechanism is decisive. Aneuploidies from random nondisjunction — trisomy 13, 18, 21 — generally recur at around 1%, with advanced maternal age raising the baseline. But if an unaffected parent carries a balanced translocation, the risk of an unbalanced, disease-causing result in the child is markedly higher and depends on parental sex: a maternal Robertsonian translocation carrier faces a 10% to 15% recurrence risk, a paternal carrier 2% to 5%.
Many common anomalies are multifactorial, blending genetic predisposition and environment, with overall recurrence usually between 3% and 5% but shifting sharply with family history, sex and severity. Neural tube defects rise from a background of about 0.3% to 4% with one affected sibling, 10% with two, and 30% to 40% if both parents are affected. Congenital heart disease runs around 3.5% for one affected sibling, 4.5% for two, and 5% to 8% if a parent is affected — higher if the mother is the affected parent, and up to 50% where the lesion belongs to an autosomal dominant syndrome such as Marfan or DiGeorge. Isolated cleft lip carries about 2.2% for a sibling (2.5% if a parent is affected); isolated cleft palate about 3.3% for a sibling. Clubfoot depends on the sex of the first affected child: 2% after an affected male, but 5% after an affected female — because clubfoot is commoner in males, an affected female implies a stronger genetic loading — rising to 25% if both a parent and a child are affected.
Two modifiers complicate every calculation. Incomplete penetrance means not everyone carrying a mutation will express it — polydactyly is a dominant condition that does not always appear. Variable expressivity means relatives with the same genotype can show very different severity, as in neurofibromatosis type 1. Because of this complexity, practice is moving toward individualised estimates: the UK’s PREGCARE (Precision Genetic Counselling and Reproduction) strategy offers couples who have had a child with a de novo mutation a personalised recurrence risk rather than a generic figure.
Practical implications at the cot-side
The through-line is sequence, not shortcut. Assess before you test: history, pedigree, examination and morphological characterisation tell you whether a genetic cause is plausible and what the highest-yield test will be. Match the test to the pattern — karyotype for a suspected common trisomy or ambiguous genitalia, and to catch a balanced translocation in recurrent miscarriage; CMA first-line for multiple unexplained anomalies or unexplained developmental delay; a gene panel only when you can name the phenotype confidently; and WES or WGS when the picture is monogenic, multi-system, or too incomplete to target. In the critically ill baby, reach for rapid WGS early, because the evidence shows it changes management and outcomes, and counsel first — trio consent, incidental and secondary findings, and the genuine possibility of a VUS or an inconclusive result. Then close the loop with a recurrence-risk conversation matched to the mechanism you found.
Key Takeaways
- Congenital anomalies affect roughly 3% to 6% of live births and are a rising cause of neonatal and infant mortality; up to 50% of infant deaths may trace to an underlying genetic disorder.
- Assessment precedes testing: history, four-generation pedigree, systematic examination and morphological characterisation decide whether — and what — to test.
- If an anomaly is confidently environmental or teratogenic, genetic testing is usually not warranted, and recurrence risk depends on re-exposure rather than inheritance.
- Match the test to the pattern: karyotype (yield ~5%) for suspected trisomies and balanced translocations; CMA (yield ~12–20%) first-line for multiple unexplained anomalies; WES/WGS (yield ~36–41%) for monogenic or complex phenotypes.
- CMA misses balanced rearrangements and raises VUS; short-read WGS misses methylation disorders, repeat expansions and balanced translocations despite its breadth.
- Rapid WGS in critically ill neonates yields diagnoses in 42–57% of cases, changing management in 30–72% and outcomes in 24–34%; even a negative result has clinical value.
- Recurrence risk follows the mechanism — Mendelian percentages, ~1% for random aneuploidy, higher for balanced-translocation carriers, and 3–5% baseline for multifactorial anomalies — modified by penetrance and expressivity.
Sources and further reading
This article is grounded in a clinical review of genetics in the newborn period. The guidelines, tools and studies it draws on include:
Mangla M, Nerakh G, Anne RP, Kaliappan A, Kaur H, Singla D. A Practical, Systematic Approach to Genetic Diagnosis in a Fetus or Neonate with Congenital Anomalies. Neoreviews. 2024 Sep 1;25(9):e537-e550. doi: 10.1542/neo.25-9-e537. PMID: 39217133.
Mangla M, Kumar N. Recurrence Risks in Congenital Anomalies: A Comprehensive Guide for Parental Counseling. Neoreviews. 2024 Dec 1;25(12):e793-e803. doi: 10.1542/neo.25-12-e793. PMID: 39616141.
American College of Medical Genetics and Genomics (ACMG) https://www.acmg.net/PDFLibrary/Exome_and_genome_sequencing_pediatric_patients.pdf
The NSIGHT1-randomized controlled trial https://pubmed.ncbi.nlm.nih.gov/29449963
The NSIGHT2-randomized controlled trial https://pubmed.ncbi.nlm.nih.gov/33157007
Effect of Whole-Genome Sequencing on the Clinical Management of Acutely Ill Infants With Suspected Genetic Disease – A randomised clinical trial https://jamanetwork.com/journals/jamapediatrics/fullarticle/2784261
Online Mendelian Inheritance in Man https://omim.org