Phenomizer - User Guide - The Human Phenotype Ontology
Contents
1. Manual Diseases 4 JORGAN ABNORMALITY gt CJ ABDOMEN ABNORMALITY gt CJ ABNORMALITY OF THE BREASTS ABNORMALITY OF THE EARS E ABNORMALITY OF THE EYES gt C ABNORMALITY OF EYE MOVEMENT 5 ABNORMALITY OF GLOBE LOCATION OR SIZE gt ANTERIOR BULGING OF THE GLOBE gt CI CRYPTOPHTHALMOS E cycioria gt DEEP SET EYES gt DHYPERTELORISM EJ HYPOTELORISM INCREASED AXIAL GLOBE LENGTH gt LARGE EYES EJ PLAGIOCEPHALY ASYMMETRY OF ORBITS gt CI SHALLOW ORBITS gt SMALL EYES gt ABNORMALITY OF REFRACTION Feature a amp category ABNORMALITY OF THE EYES 1 Item Modifier HP 0000494 B category GROWTH ABNORMALITY 1 Item DOWNWARD SLANTING PALPEB observed HP 0001509 SHORT STATURE category HEAD AND NECK ABNORMALITY 2 Items observed HP 0000494 HP 0002162 LOW POSTERIOR HAIRLINE show OMIM entries display in ontology Y INTEGUMENT ABNORMALITY 2 Items DOWNWARD SLANTING PALPEB observed observed CAFE AU LAIT SPOTS LOW POSTERIOR HAIRLINE add to patient s features gt JABNORMALITY OF TEAR GLANDS OR TEAR PRODUCTION gt ABNORMALITY OF THE ANTERIOR SEGMENT OF THE EYE ABNORMALITY OF THE CONJUNCTIVA gt ABNORMALITY OF THE EYELID observed observed Mode of inheritance Figure 3 You can browse through the Human Phenotype Ontology in a top down way to find the most specific term describing your patient s phenoty2. 0 0 0000004 5 3 3 2 Specific scoring of suggestive HPO terms 0 0 0 000000004 5 3 4 Symmetric and Asymmetric Searches oaoa 5 4 Export search results 6 1 Disclaimer The Phenomizer is intended to be used by qualified and licensed physicians in order to provide assistance in reaching the correct diagnosis in patients with hereditary diseases and for use as a teaching aid The Phenomizer does not make diagnoses Rather it produces a ranked list of possibilities that can be used by physicians as a part of the diagnostic workup The Phenomizer does not contain information about all possible diagnoses or even all possible hereditary diseases The Phenomizer should not be used to make medical decisions without the advice of a physician Phenomizer Manual 2 How to define the query The medical terms accessible to the Phenomizer are based on the terminology of the Human Phenotype On tology HPO The Phenomizer offers three different approaches to find the appropriate term for a phenotypic abnormality indicated by the three tabs on the left hand side Feature Disease and Ontology 2 1 Entering clinical data of a patient Under the Feature tab you can search for the appropriate HPO terms by typing part of the name synonym or HPO identifier into the search box This search box provides autocompletion functionality Possible matches are shown below while the query is typed in After pressing the search button features of the H
3. PO that best agree with the query are listed in a table For every feature displayed in the table you can open a context menu by right clicking on the row This will offer you further information about the specific HPO term Figure 1 An HPO term can be added to the Patient s features list by double clicking drag and drop or via the context menu In our example we selected the phenotypic features short stature low posterior hair line and downward slanting palpebral fissures Features Diseases Ontology HPO id Feature HP 0008929 ASYMMETRIC SHORT STATURE HP 0003522 DISPROPORTIONATE SHORT STATURE SHORT TRUN HP 0008903 MESOMELIC SHORT STATURE HP 0003502 MILD SHORT STATURE HP 0003503 MILD TO MODERATE SHORT STATURE HP 0008848 MODERATELY SHORT STATURE HP 0003501 POSTNATAL SHORT STATURE HP 0008913 PRE AND POSTNATAL SHORT STATURE HP 0009885 PRENATAL SHORT STATURE Open a new window where all diseases annotated to g HP 0003508 PROPORTIONATE SHORT STATURE ie short stature are listed HP 0008879 RELATIVE SHORT STATURE HP 0008905 RHIZOMELIC SHORT STATURE HP 0008920 SEVERE SHORT STATURE POSTNATAL ONSET Displays a draggable image where the HPO term and all HP 0008900 SEVERELY DISPROPORTIONATE SHORT STATURE zits ancestors are displayed HP 0001509 SHORT STATURE S Displays this H PO term and ii d Page fi of2 gt bi S all its descendants A Add s this HPO term to the list of feature
4. Phenomizer User Guide Sebastian Kohler Peter Krawitz Peter N Robinson Abstract In this manual we will describe the features of the Phenomizer and guide the user through the workflow of using the Phenomizer for the differential diagnosis In the first section we illustrate how to find and enter the clinical features of a patient and will explain the different methods for performing ontological searches with the Phenomizer In the second section we will show how to use the Phenomizer to provide suggestions for continuing the differential diagnostic process by identifying clinical features that can best differentiate among the top candidate differential diagnoses Finally we show in the last section how search results may be exported for documentation and or further analysis Contents 1 Disclaimer 1 2 How to define the query 1 2 1 Entering clinical data of a patient oaa aaa ee 1 2 2 Patient s Features a mene m GA Bg ee ee ea a Big ae ee A a a ae 2 2 2b Remove features e s anar oats a Dele eae eee owe Oe Pe eM le ee eo 2 22 2 Retine query airm haps nia me ae he oe Se Gi ie hie eee Gia uals 2 3 Get differential diagnoses 4 Bul Settings oso ergs t Ra A ei Se ee Sa Ge aed iy ee ee Sea See ae ty oh ee Be 4 3 2 The first differential diagnosis 2 2 2 2 0 2 02 eee eee ee 4 3 3 Improve the differential diagnosis 2 0 0 00 0002 pe ee 5 3 3 1 Binary scoring of suggestive HPO terms 0 0 0
5. T SPOTS MULTIPLE HP 0007718 KEY HOLE IRIS RETINA CHOROID COLOBOMA 7 0 2096 LEOPARD SYNDROME 1 HP 0002419 MOLAR TOOTH SIGN ON MRI E 0 2096 CRANIOFRONTONASAL SYNDROME HP 0007438 MOTTLED PIGMENTATION OF THE TRUNK AND PROXI J 0 2096 NOONAN SYNDROME 1 Page t of 591 gt Dl Displaying Features 1 15 of 8855 4 fr ofsia Ka ny ISi Tor Show OMIM entry Display overlap Figure 5 For each of the syndrome in the result list there exists a context menu from which allows you e g to display the overlap between the patients features and the syndrome s features In this example Low posterior hairline and the other red terms are shared by the disease and the query Lypmhedema and the other yellow terms are used to annotate the disease but were not among the query terms The blue term Cafe au lait spots was used in the query but is not used to annotate the disease 3 3 Improve the differential diagnosis The Phenomizer offers the possibility to improve your differential diagnoses You can select several at least two of the listed syndromes by marking the corresponding checkbox By clicking the Improve differential diagnosis button the Phenomizer will generate a list of phenotypic features that are annotated only to a subset of the selected diseases binary search or that are specific for one syndrome specific search 3 3 1 Binary scoring of suggestive HPO terms The binary scoring method prioritizes terms that are used to anno
6. ay their annotated phenotypic features Features that are relevant in your case can be added to the Patient s features If your patient shares some phenotypic abnormalities with a different genetic syndrome you are familiar with the Disease tab offers a quick way to select appropriate features Using this tab you can use the search box to search for syndromes by entering a part of their name or identifier The context menu right click shows a list of all the features used to describe the syndrome You can also use the hyperlink to open a new window with the OMIM entry for this syndrome Appropriate phenotypic features can now be added to the Patient s features by double clicking or drag and drop Figure 2 In the example we have selected Cafe au lait spots a clinical feature in Fanconi anemia MIM 227650 You may also browse through the HPO under the Ontology tab to search for the most specific term that describes your patient s phenotypic feature In an ontology child terms have the same or higher information content compared to their parents and will thus make a query more specific Again every feature has its own context menu and can be added to the list of the Patient s features as described above In our example we chose hypertelorism as a precise description of the abnormality of the eyes found in our patient Figure 3 2 2 Patient s Features The features that have been selected as above are now listed on the right hand side of
7. d clinical features are being sought to further support the diagnosis The methods starts by taking all terms excluding ancestors of the selected syndromes We then remove all terms that were implicitly listed in the query i e the query terms and their descendents For each of the terms we simply take the number of selected syndromes showing this feature If multiple terms were used to annotate the same amount of selected syndromes we prioritize the features being most specific highest information content 3 4 Symmetric and Asymmetric Searches Using the standard search method described in the main manuscript only the terms of the diseases that best match the query are used for calculating the score This may not be optimal for cases in which a patient is found Page 5 Phenomizer Manual to have only a few clinical features but a disease is characterized by numerous signs and symptoms In this case a symmetric version of the score may be used ti sim Q gt D I avg max IC t1 t2 2 i co ER 1 jag pag Fett t2ED This equation differs from the one presented in the main text by the fact that it also searches for the best match in the query terms for each of the terms annotating a disease The Phenomizer automatically opens an option panel for switching to symmetric searches once 6 or more search terms have been entered 4 Export search results The results of the phenotypic analysis can be exported as a csv
8. ed in OMIM Physicians may then want to use the Improve differential diagnosis function to obtain a list of clinical features that may help with the further differential diagnostic process see section 3 3 Patient s Features Diagnosis Algorithm resnik Unsymmetric No grasm 6 Features i OMIM name Genes 0478 i DERMOID CYSTS FAMILIAL FRONTONASAL 1195 HYPERTELORISM 1593 i GASTROCUTANEOUS SYNDROME 1673 i RUSSELL SILVER SYNDROME X LINKED 2549 i GRANDDAD SYNDROME 2549 i EPICANTHUS 3117 RUBINSTEIN TAY BI SYNDROME EP300 CREBBP 3117 i BRANCHIOOCULOFACIAL SYNDROME TFAP24 3117 i MOVAT VVILSON SYNDROME ZEB2 3117 i BANNAYAN RILEY RUVALCABA SYNDROME PTEN 3117 LEOPARD SYNDROME 1 3117 i CEREBRAL DYSGENESIS NEUROPATHY ICHTHYOSIS AND P 3117 i NOONAN LIKEMULTIPLE GIANT CELL LESION SYNDROME 3117 i MOMO SYNDROME 3117 XERODERMA PIGMENTOSUM COMPLEMENTATION GROUP B zi Page fi of 160 gt Pi Improve Differential Diagnosis Download Results Selection for i Improve Differential Diagnosis Figure 4 The Phenomizer ranks the diagnoses by their P value For every row you can use the context menu right click and get further information about the suggested diagnosis e g links to the genes known to cause this syndrome when mutated or you can get a visualization of the overlap between the query and a specific syndrome in the context of the HPO structure where the query terms are displayed in blue the terms be
9. file or PDF file The csv file comma separated values can be imported into spreadsheet programs such as OpenOffice of Microsoft Excel The PDF file can be used to document the differential diagnostic process in the patient chart Fields are provided to enter the patient s name and other information A summary of the query terms and the ranked differential diagnoses is given Finally for every syndrome of the ranking a list of phenotypic features that are specific for this syndrome among all the diseases in the differential diagnosis is provided Page 6
10. longing to the disease in yellow and terms that are shared by query and disease in red see Figure 5 Note that you you can drag the displayed image inside the popup window Page 4 3 3 Improve the differential diagnosis Phenomizer Manual Menu Help Features Diseases Ontology Patient s Features Diagnosis Enter feature Algorithm resnik Unsymmetric No grasm 5 Features drogable mons HPO id Feature p value a OMIM name gt HP 0003013 BULGING EPIPHYSES T 04338 NOONAN LIKEMULTIPLE GIANT CELL LESION SYI HP 0000893 BULGING OF THE COSTOCHONDRAL JUNCTION F 01338 HYPERTELORISM TEEBI TYPE HP 0000186 CARP LIKE MOUTH 04338 GASTROCUTANEOUS SYNDROME l HP 0006170 CHESS PAWIN DISTAL PHALANGES 01338 WIDOWS PEAK HP 0006168 CHUBBY PROXIMAL PHALANGES 0 1338 RUSSELL SILVER SYNDROME X LINKED HP 0006665 COAT HANGER DEFORMITY OF LOWER RIBS E 04434 FACIAL DYSMORPHISM WITH MULTIPLE MALFOR HP 0003205 CURVILINEAR PROFILES ULTRASTRUCTURALLY 01434 ALDOLASE A FRUCTOSE BISPHOSPHATE HP 0000295 DOLL LIKE FACIES 01434 PTERYGIUM COLLI ISOLATED HP 0006129 DRUMSTICK TERMINAL PHALANGES E 01434 HYPERTELORISM HP 0006422 ERLENMEYER FLASK DEFORMITY OF DISTAL FEMUR 0 1960 CATARACT ABERRANT ORAL FRENULA AND G HP 0004975 ERLENMEYER FLASK DEFORMITY OF THE FEMURS E 01992 BRANCHIAL ARCH SYNDROME X LINKED HP 0003208 FINGERPRINT PROFILES ULTRASTRUCTURALLY 0 1992 CAFE AU LAI
11. pic feature Page 3 Phenomizer Manual 3 Get differential diagnoses 3 1 Settings Phenotypic similarity can be calculated in a one sided asymmetric or two sided symmetric fashion see sec tion 3 4 In general we recommend starting with an asymmetric search which is the default However using symmetric searches may better distinguish between diseases with few and many clinical features This can be done using the settings menu on the upper left You may also select a particular multiple testing procedure from the menu The default setting is the Benjamini Hochberg procedure which is recommended for normal users 3 2 The first differential diagnosis Click the Get diagnosis button to cause the Phenomizer to compute a list of differential diagnoses that are ranked by their P value Figure 4 A significant P value does not mean that the diagnosis is confirmed We suggest that it can be taken to mean that the diagnosis is at least plausible and that the physician should consider statistically significant differential diagnoses carefully On the other hand if no disease is found to have a significant P value for the features entered by the physician we would interpret this to mean that the combination of clinical features is not per se sufficient to make a diagnosis It may mean that the disease which the patient has is not in the database being used by the program which currently encompasses the mainly Mendelian diseases list
12. s that the patient shows Figure 1 The context menu of an HPO term allows you to add the feature to the list of Patient s features You may also gather further information about the term by listing all OMIM diseases that are annotated with this feature or by displaying the feature in the context of the structure of the Human Phenotype Ontology http www human phenotype ontology org Page 1 2 2 Patient s Features Features Diseases Ontology Patient s Features FANCONI ANEMIA HPO Feature a Modifier OMIM id OMM entry category ABNORMALITY OF THE EYES 1 Item zie Rata tenev a Sa HP 0000494 DOWNWARD SLANTING PALPEB observed 227645 FANCONI ANEMIA 3 v 227646 FANCONI ANEMIA Cor Show OMIM entry THESES for FANCONI ANEMIA eaa g x f 600901 FANCONI ANEMIA COMPLEMENTATION GROUP E pai es paced HP 0003974 ABSENT OSSIFICATION ABSENCE OF RADIUS f HP 0001017 ANEMIC PALLOR pa HP 0009777 APLASIA OF THE THUMB pervert HP 0001892 BLEEDING DIATHESIS HP 0000957 CAFE AU LAIT SPOTS served HP 0002564 CARDIAC MALFORMATION HP 0003221 CHROMOSOMAL BREAKAGE INDUCED BY DIE HP 0000028 CRYPTORCHIDISM HP 0000404 DEAFNESS HP 0003213 DEFICIENT EXCISION OF U V INDUCED PYRIMIDI HP 0000081 DUPLICATED COLLECTING SYSTEM zl Page l of1 gt i OMIM entries 1 Forty J Clear Mode oF inheritance 2 Y Get diagnosis Figure 2 You can use the search box under the disease tab to look for syndromes and displ
13. tate half of the selected differential diagnoses and are not present in the query The method starts by generating a list of all terms including ancestors of the selected syndromes From this list all terms that already have been included in the query patient s features and their respective ancestors are removed We thus have a set of phenotypic features that are only used to annotate the set of selected candidate syndromes and have not already been used as query term For each of these terms we calculate a score as the fraction of selected syndromes that show the particular feature compared to all selected syndromes If multiple syndromes have the same score we favor terms that rather unspecific meaning that we prioritize features that have a low information content This feature can be used to identify general classes of phenotypic features that will best differentiate between about half of the marked differential diagnoses and may help to narrow down the differential diagnosis Users can right click on one of the proposal terms to obtain a submenu with more specific terms 3 3 2 Specific scoring of suggestive HPO terms The specific scoring method prioritizes terms that are used to annotate a minimal number of the selected differential diagnoses and are not present in the query This function can be more useful than the Binary search function if there is some reason to believe that one of the top differential diagnoses is the correct diagnosis an
14. the Phenomizer window In general it is best to enter as many of the clinical features of a patient as possible in order to improve the specificity of the differential diagnoses proposed by the Phenomizer However it is also possible to add features one by one to explore the phenotypic space or to learn about the diseases that are characterized by sets of phenotypic features 2 2 1 Remove features The Patient s features are listed on the right hand side To delete a feature from this list right click on the particular row and select remove from the context menu To delete all features from the list you can use clear button below the table 2 2 2 Refine query You can choose for every feature whether it should be regarded as observed or mandatory In the default setting observed the similarities between the patient and particular syndromes are calculated as described in the paper If a term is denoted as mandatory all syndromes not having this particular feature or a descendent thereof will not be regarded as a possible differential diagnosis and are thus filtered out You can also select the mode of inheritance for the syndrome observed in your patient if it is suspected because of the pedigree Again all syndromes that do not exhibit this mode of inheritance are not regarded as a possible differential diagnoses and are filtered out Phenomizer Manual Page 2 2 2 Patient s Features Phenomizer
Download Pdf Manuals
Related Search