Merge branch 'master' of https://github.com/hartwigmedical/hmftools

hmf-common/src/main/resources/actionability/primary_tumor_locations_mapping.csv

@@ -1,13 +1,14 @@
 primaryTumorLocation,doids
 Adrenal,3953
+Anus,14110
 Biliary,4607;4947
-Bone/Soft tissue,201;1115;3347
+Bone/Soft tissue,201;1115;3347; 9253
 Breast,1612
-Colon/Rectum,9256;219
-CUP,UNMAPPED
+Colon/Rectum,9256
 Double primary,UNMAPPED
 Esophagus,5041;4944
 Eye,2174
+GI-tract,3119
 Head and neck,11934
 Kidney,4450;263
 Liver,3571
@@ -16,7 +17,7 @@ Lymphoid,60058
 Mesothelioma,1790
 Myeloid,70004
 Nervous system,3093;1319
-NET,169;1800;4434;1798;5410
+Neuroendocrine,169;1800;4434;1798;5410;0050872
 Ovary,2394
 Pancreas,1793
 Penile,11615
@@ -30,4 +31,6 @@ Thyroid,1781
 Unknown,UNMAPPED
 Urinary tract,3996
 Uterus,363;1380
-Vulva,1245;119
+Vagina,119
+Vascular,175
+Vulva,1245

knowledgebase-generator/src/main/java/com/hartwig/hmftools/knowledgebasegenerator/ViccAmpsDelExtractorTestApplication.java

@@ -28,7 +28,7 @@ public class ViccAmpsDelExtractorTestApplication {
     public static void main(String[] args) throws IOException, InterruptedException {
         String viccJsonPath = System.getProperty("user.home") + "/hmf/projects/vicc/all.json";
 
-        String source = "oncokb";
+        String source = "civic";
         LOGGER.info("Reading VICC json from {} with source '{}'", viccJsonPath, source);
         List<ViccEntry> viccEntries = ViccJsonReader.readSingleKnowledgebase(viccJsonPath, source);
         LOGGER.info("Read {} entries", viccEntries.size());
@@ -52,7 +52,7 @@ public static void main(String[] args) throws IOException, InterruptedException
             List<EventType> eventType = EventTypeAnalyzer.determineEventType(viccEntry);
 
             for (EventType type : eventType) {
-              //  LOGGER.info("gene: " + type.gene() + " name: " + type.name() + " eventMap: " + type.eventMap() + " source: " + type.source());
+                LOGGER.info("gene: " + type.gene() + " name: " + type.name() + " eventMap: " + type.eventMap() + " source: " + type.source());
                 // Generating known events
                 //TODO: map every genomic event to one object
                 //TODO: if combined event use single event for determine known events

knowledgebase-generator/src/main/java/com/hartwig/hmftools/knowledgebasegenerator/eventtype/EventTypeAnalyzer.java

@@ -53,7 +53,6 @@ public static List<EventType> determineEventType(@NotNull ViccEntry viccEntry) {
                     }
 
                     eventMap.put(gene, Lists.newArrayList(name));
-                    LOGGER.info(eventMap);
 
                     if (eventMap.isEmpty()) {
                         LOGGER.warn("Skipping feature interpretation of '{}' on gene '{}' with biomarker type '{}' on source '{}' ",
@@ -110,16 +109,15 @@ public static List<EventType> determineEventType(@NotNull ViccEntry viccEntry) {
                             if (name.split(" ", 2).length == 1 && gene.contains("-")) {
                                 eventInfo = FUSION_PAIR;
                                 eventMap.put(gene, Lists.newArrayList(eventInfo));
-                                LOGGER.info(eventMap);
                             } else {
                                 eventInfo = name.split(" ", 2)[1];
-                                if (eventInfo.contains("fusion") || eventInfo.contains("Fusion") ) {
+                                if (eventInfo.contains("fusion") || eventInfo.contains("Fusion")) {
                                     if (gene.contains("-")) {
                                         eventInfo = FUSION_PAIR;
                                     } else {
                                         eventInfo = FUSION_PROMISCUOUS;
                                     }
-                              }
+                                }
 
                                 if (eventInfo.equals(".")) {
                                     eventInfo = ONCOGENIC_MUTATION;
@@ -165,22 +163,25 @@ public static List<EventType> determineEventType(@NotNull ViccEntry viccEntry) {
                                 combinedEvent = true;
 
                                 if (eventMap.size() == 0) {
-                                    eventMap.put(fusion, Lists.newArrayList("Fusion"));
+                                    eventMap.put(fusion, Lists.newArrayList(FUSION_PAIR));
                                     if (eventMap.containsKey(geneVariant)) {
-                                        eventMap.put(geneVariant, Lists.newArrayList("Fusion", variant));
+                                        eventMap.put(geneVariant, Lists.newArrayList(FUSION_PAIR, variant));
                                     } else {
-                                        eventMap.put(fusion, Lists.newArrayList("Fusion"));
+                                        eventMap.put(fusion, Lists.newArrayList(FUSION_PAIR));
                                         eventMap.put(geneVariant, Lists.newArrayList(variant));
                                     }
                                 }
                             } else if (combinedEventConvertToSingleEvent.length >= 2) {
                                 LOGGER.warn("This event has more events, which is not interpretated!");
                             }
                         } else if (name.contains("-")) {
-                            eventMap.put(name, Lists.newArrayList("Fusion"));
+                            eventMap.put(name, Lists.newArrayList(FUSION_PAIR));
+                        } else if (name.equals("TRUNCATING FUSION")) {
+                            eventMap.put(gene, Lists.newArrayList(name));
+                        } else if (name.contains("FUSION") || name.contains("FUSIONS")) {
+                            eventMap.put(gene, Lists.newArrayList(FUSION_PROMISCUOUS));
                         } else {
                             if (name.contains("+")) {
-                                LOGGER.info("combined: " + name);
 
                                 combinedEvent = true;
                                 String[] combinedEventConvertToSingleEvent = name.replace("+", " ").split(" ", 2);
@@ -201,6 +202,14 @@ public static List<EventType> determineEventType(@NotNull ViccEntry viccEntry) {
                                 eventMap.put(gene, Lists.newArrayList(name));
                             }
                         }
+                    } else if (name.contains("+") && !name.contains("c.") && !name.contains("C.")) {
+                        combinedEvent = true;
+                        String[] combinedEventConvertToSingleEvent = name.split("\\+", 2);
+                        String event1 = combinedEventConvertToSingleEvent[0];
+                        String event2 = combinedEventConvertToSingleEvent[1];
+
+                        eventMap.put(gene, Lists.newArrayList(event1, event2));
+
                     } else {
                         eventMap.put(gene, Lists.newArrayList(name));
                     }

patient-db/src/main/resources/tumor_location_mapping.csv

@@ -77,7 +77,7 @@ searchTerm,primaryTumorLocation,subType
 "Gallbladder cancer",Biliary,"Gall bladder"
 "Gasto Intestinal Stroma Tumor (GIST)",Bone/Soft tissue,"Undifferentiated sarcoma"
 "gastric carcinoma",Stomach,"Gastric"
-"gastrinoma in stomach",NET,"Gastrinoma"
+"gastrinoma in stomach",Neuroendocrine,"Gastrinoma: Stomach"
 "Gastrointestinal cancer/pancreatic-biliary type",Pancreas,"Biliary type"
 "Gastrointestinal Carcinoid Tumor",Bone/Soft tissue,"Gastrointestinal Carcinoid"
 "Gastrointestinal Stromal Tumors (GIST)",Bone/Soft tissue,"Gastrointestinal stromal tumor (GIST)"
@@ -118,12 +118,12 @@ searchTerm,primaryTumorLocation,subType
 "jejunum carcinoma",Small intestine,"Jejunum"
 "kidney",Kidney,""
 "Kidneycell carcinoma",Kidney,"Renal cell"
-"lage cell neuroendocrien carcinoma",NET,"Large cell"
-"large cell neuroendocrien carcinoma",NET,"Large cell"
+"lage cell neuroendocrien carcinoma",Neuroendocrine,"NEC: Large cell"
+"large cell neuroendocrien carcinoma",Neuroendocrine,"NEC: Large cell"
 "Larynxcarcinoma",Head and neck,"Laryngeal"
 "left tuba",Ovary,"Left tuba"
 "Leiomyosarcoom",Bone/Soft tissue,"Leiomyosarcoma"
-"Long-NET",NET,"Lung"
+"Long-NET",Neuroendocrine,"NET: Lung"
 "Lung Cancer: Non-Small Cell",Lung,"Non-Small Cell"
 "Lung Cancer: Non-Small Cell_",Lung,"Non-Small Cell"
 "Lung Cancer: Small Cell",Lung,"Small Cell"
@@ -134,7 +134,7 @@ searchTerm,primaryTumorLocation,subType
 "Lymphoma: Non Hodgkin",Lymphoid,"Non-Hodgkin"
 "Lymphoma: other",Lymphoid,"Other"
 "Malignant peripheral nerve sheath tumor",Bone/Soft tissue,"Neurofibrosarcoma"
-"maligne feochromocytoom",NET,"Adrenal"
+"maligne feochromocytoom",Neuroendocrine,"NET: Adrenal"
 "Medulloblastoma",Nervous system,"Medulloblastoma"
 "Melanoma",Skin,"Melanoma"
 "Melanoma_",Skin,"Melanoma"
@@ -154,34 +154,34 @@ searchTerm,primaryTumorLocation,subType
 "multipel myeloma",Myeloid,"Multiple Myeloma"
 "myo-epithelioma distale external auditory canal",Head and neck,"External auditory canal"
 "Nasal Cavity and Paranasal Sinus Cancer",Head and neck,"Nasal cavity and paranasal sinuses"
-"NEC from pancreatic",NET,"Pancreatic NEC"
-"NET colon ascendens",NET,"Colon/Rectum"
-"NET of unknown origin",NET,""
-"NET prostate",NET,"Prostate"
-"NET rectosigmoid",NET,"Rectosigmoid"
-"NET unknown primary",NET,""
-"NET unknown primary location",NET,""
-"NET: Large cell neuroendocrine carcinoma (LCNEC)",NET,"Large cell NEC"
-"NET: Lung",NET,"Lung"
-"NET: lung or small intestine",NET,"Lung or small intestine"
-"NET: Lung_",NET,"Lung"
-"NET: other",NET,""
-"NET: Other (stomach)",NET,"Stomach"
-"NET: Pancreatic",NET,"Pancreatic"
-"NET: Pancreatic_",NET,"Pancreatic"
-"NET: Small Intestinal",NET,"Small Intestinal"
-"Neuro endocrine carcinoma",NET,"NEC"
-"Neuro endocrine carcinoma (NEC), primary unknown",NET,"NEC"
-"Neuro endocrine tumor",NET,""
-"Neuro-endocrinal cervix carcinoma",NET,"Cervix"
-"Neuro-endocrine carcinoma",NET,"NEC"
-"Neuro-endocrine carcinoma of the bladder",NET,"Bladder NEC"
-"Neuro-endocrine carcinoma with unknown primary tumor",NET,"NEC"
-"neuro-endocrine tumor, possible transformation of prostatecarcinoma",NET,"Possible transformation of prostatecarcinoma"
-"Neuro-endorine tumor",NET,""
+"NEC from pancreatic",Neuroendocrine,"NEC: Pancreatic"
+"NET colon ascendens",Neuroendocrine,"NET: Colon"
+"NET of unknown origin",Neuroendocrine,"NET"
+"NET prostate",Neuroendocrine,"NET: Prostate"
+"NET rectosigmoid",Neuroendocrine,"NET: Rectosigmoid"
+"NET unknown primary",Neuroendocrine,"NET"
+"NET unknown primary location",Neuroendocrine,"NET"
+"NET: Large cell neuroendocrine carcinoma (LCNEC)",Neuroendocrine,"NEC: Large cell"
+"NET: Lung",Neuroendocrine,"NET: Lung"
+"NET: lung or small intestine",Neuroendocrine,"NET: Lung or small intestine"
+"NET: Lung_",Neuroendocrine,"NET: Lung"
+"NET: other",Neuroendocrine,"NET"
+"NET: Other (stomach)",Neuroendocrine,"NET: Stomach"
+"NET: Pancreatic",Neuroendocrine,"NET: Pancreatic"
+"NET: Pancreatic_",Neuroendocrine,"NET: Pancreatic"
+"NET: Small Intestinal",Neuroendocrine,"NET: Small Intestinal"
+"Neuro endocrine carcinoma",Neuroendocrine,"NEC"
+"Neuro endocrine carcinoma (NEC), primary unknown",Neuroendocrine,"NEC"
+"Neuro endocrine tumor",Neuroendocrine,"NET"
+"Neuro-endocrinal cervix carcinoma",Neuroendocrine,"NEC: Cervical"
+"Neuro-endocrine carcinoma",Neuroendocrine,"NEC"
+"Neuro-endocrine carcinoma of the bladder",Neuroendocrine,"NEC: Bladder"
+"Neuro-endocrine carcinoma with unknown primary tumor",Neuroendocrine,"NEC"
+"neuro-endocrine tumor, possible transformation of prostatecarcinoma",Neuroendocrine,"NET: Possible transformation of prostatecarcinoma"
+"Neuro-endorine tumor",Neuroendocrine,"NET"
 "Neuroblastoma",Nervous system,"Neuroblastoma"
-"neuroendocrine carcinoma",NET,"NEC"
-"Neuroendocrine tumor",NET,""
+"neuroendocrine carcinoma",Neuroendocrine,"NEC"
+"Neuroendocrine tumor",Neuroendocrine,"NET"
 "non-HIV Kaposisarcoom",Skin,"Other"
 "Non-small cell carcinoma NOS (mostly resembling lung carcinoma): working diagnosis ""lung carcinoma""",Lung,"Non-Small Cell"
 "Not in dropdown list (please specify below)_adenocarcinoom van de klier van Bartholin vulva links",Vulva,"Adenocarcinoma"
@@ -200,12 +200,12 @@ searchTerm,primaryTumorLocation,subType
 "Other (please specify below)_empithelioid hemangioendothelioom",Vascular,"Hemangioendothelioma"
 "Other (please specify below)_empithelioid hemangioendothelioom",Vascular,"Hemangioendothelioma"
 "Other (please specify below)_Esophagus squamous cell carcinoma (ESCC)",Esophagus,""
-"Other (please specify below)_LCNEC",NET,"Large cell NEC"
+"Other (please specify below)_LCNEC",Neuroendocrine,"NEC: Large cell"
 "Other (please specify below)_Merkel Cel Carcinoom",Skin,"Merkel-cell carcinoma"
 "Other (please specify below)_myo-epitheliaal carcinoom",Bone/Soft tissue,"Undifferentiated sarcoma"
-"Other (please specify below)_NEC",NET,"NEC"
-"Other (please specify below)_NEC Colon",NET,"Colon NEC"
-"Other (please specify below)_NET; Rectal",NET,"Rectum"
+"Other (please specify below)_NEC",Neuroendocrine,"NEC"
+"Other (please specify below)_NEC Colon",Neuroendocrine,"NEC: Colon"
+"Other (please specify below)_NET; Rectal",Neuroendocrine,"NET: Rectal"
 "Other (please specify below)_ovariele stromale tumor",Ovary,""
 "Other (please specify below)_pleiomorf",Bone/Soft tissue,"Pleomorphic Sarcoma"
 "Other (please specify below)_Trichilemmal carcinoma ear",Skin,"Trichilemmal carcinoma"
@@ -215,8 +215,8 @@ searchTerm,primaryTumorLocation,subType
 "Other (please specify below)_Vaginal cancer",Vagina,""
 "Ovarian cancer",Ovary,""
 "Pancreatic cancer",Pancreas,""
-"Pancreatic Neuroendocrine Tumors (Islet Cell Tumors)",NET,"Pancreatic"
-"Paraganglioma / NET lung?",NET,"Paraganglioma (lung?)"
+"Pancreatic Neuroendocrine Tumors (Islet Cell Tumors)",Neuroendocrine,"NET: Pancreatic"
+"Paraganglioma / NET lung?",Neuroendocrine,"NET: Paraganglioma (lung?)"
 "Penile cancer",Penile,""
 "plexuspapilloom",Nervous system,"Plexuspapilloom"
 "primary unknown",Unknown,"Unknown"
@@ -225,7 +225,7 @@ searchTerm,primaryTumorLocation,subType
 "Pseudomyxoma peritonei",Colon/Rectum,"Low-grade mucinous adenocarcinoma"
 "Rectum",Colon/Rectum,"Rectum"
 "Rectum carcinoma",Colon/Rectum,"Rectum"
-"Rectum NET",NET,"Rectum"
+"Rectum NET",Neuroendocrine,"NET: Rectal"
 "rectumcarcinoom",Colon/Rectum,""
 "renal carcinoma",Kidney,"Renal cell"
 "Renal Cell Cancer",Kidney,"Renal cell"
@@ -275,9 +275,9 @@ searchTerm,primaryTumorLocation,subType
 "Synovial carcinoma",Bone/Soft tissue,"Synovial"
 "T-cell lymphoma",Lymphoid,"T-cell"
 "Testicular cancer",Testis,""
-"thymic NET",NET,"Thymus"
+"thymic NET",Neuroendocrine,"NET: Thymus"
 "Thymoma and Thymic Carcinoma",Thymus,""
-"thymus-neuroendocrine tumor",NET,"Thymus"
+"thymus-neuroendocrine tumor",Neuroendocrine,"NET: Thymus"
 "Thyroid Cancer",Thyroid,""
 "Thyroid Cancer_",Thyroid,""
 "Tongue carcinoma lateral side",Head and neck,"Tongue"
@@ -290,6 +290,7 @@ searchTerm,primaryTumorLocation,subType
 "unknow primary tumor",Unknown,"Unknown"
 "unknown",Unknown,"Unknown"
 "Unknown primary",Unknown,"Unknown"
+"Unknown primary location melanoma",Unknown,"Melanoma"
 "unknown primary tumor in tractus genitalis",Unknown,"Tractus Genitalis"
 "Unknown, probably head and neck cancer",Unknown,"Possibly Head and neck"
 "Urachuscarcinoma",Urinary tract,"Urachal carcinoma"

purity-ploidy-estimator/src/main/resources/circos/input.template

@@ -37,7 +37,7 @@ show_tick_labels    = no
 	<<include etc/image.conf>>
 </image>
 
-karyotype = data/karyotype/karyotype.human.txt
+karyotype = KARYOTYPE
 
 chromosomes_units           = 1000000
 chromosomes_display_default = yes

sage/README2.md

@@ -365,20 +365,19 @@ Peak memory is measure in gigabytes.
 
 Threads | Elapsed Time| CPU Time | Peak Mem
 ---|---|---|---
-1 | 597 | 601 | 44
-8 | 87 | 653 | 83
-16 | 50 | 722 | 83 
-24 | 38 | 773 | 82 
-32 | 32 | 830 | 84
-48 | 30 | 1150 | 82
-64 | 29 | 1478 | 82
-72 | 30 | 1660 | 82
+1 | 529 | 539 | 55
+8 | 77 | 594 | 68
+16 | 46 | 675 | 68 
+24 | 32 | 671 | 67 
+32 | 28 | 748 | 67
+48 | 27 | 1047 | 66
 
  ## Version History
  - Upcoming
    - Multiple tumor support
    - Multiple reference (or RNA) support
    - Removed explicit RNA support (can use additional reference instead)
+   - Performance and memory improvements
  - [2.1](https://github.com/hartwigmedical/hmftools/releases/tag/sage-v2.1)
    - Reduced memory footprint
    - Add version info to VCF