feat: restore gene model url (#177)

galaxyproject · Nov 17, 2024 · 94f19e4 · 94f19e4
1 parent 4f1f59f
commit 94f19e4
Show file tree

Hide file tree

Showing 10 changed files with 86 additions and 30 deletions.
diff --git a/app/apis/catalog/brc-analytics-catalog/common/entities.ts b/app/apis/catalog/brc-analytics-catalog/common/entities.ts
@@ -15,6 +15,7 @@ export interface BRCDataCatalogGenome {
   chromosomes: number | null;
   coverage: string | null;
   gcPercent: number;
+  geneModelUrl: string | null;
   isRef: boolean;
   length: number;
   level: string;

diff --git a/app/components/Entity/components/AnalysisMethod/types.ts b/app/components/Entity/components/AnalysisMethod/types.ts
@@ -4,6 +4,6 @@ import { ANALYSIS_METHOD } from "../../../../apis/catalog/brc-analytics-catalog/
 export interface Props {
   analysisMethod: ANALYSIS_METHOD;
   content: ReactNode;
-  geneModelUrl: string;
+  geneModelUrl: string | null;
   genomeVersionAssemblyId: string;
 }
diff --git a/app/utils/galaxy-api.ts b/app/utils/galaxy-api.ts
@@ -32,7 +32,7 @@ const WORKFLOW_LANDING_URL_PREFIX =
 export async function getWorkflowLandingUrl(
   workflowId: WORKFLOW_ID,
   referenceGenome: string,
-  geneModelUrl: string
+  geneModelUrl: string | null
 ): Promise<string> {
   const body: WorkflowLandingsBody = {
     public: true,
@@ -74,7 +74,7 @@ function buildFastaUrl(identifier: string): string {
 function getWorkflowLandingsRequestState(
   workflowId: WORKFLOW_ID,
   referenceGenome: string,
-  geneModelUrl: string
+  geneModelUrl: string | null
 ): WorkflowLandingsBodyRequestState {
   if (workflowId === WORKFLOW_ID.VARIANT_CALLING && geneModelUrl) {
     return {

diff --git a/app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts b/app/viewModelBuilders/catalog/brc-analytics-catalog/common/viewModelBuilders.ts
@@ -231,7 +231,7 @@ export const buildGenomeAnalysisMethod = (
 ): ComponentProps<typeof C.AnalysisMethod> => {
   return {
     ...analysisMethodProps,
-    geneModelUrl: "",
+    geneModelUrl: genome.geneModelUrl,
     genomeVersionAssemblyId: genome.accession,
   };
 };

diff --git a/files/build-catalog.ts b/files/build-catalog.ts
@@ -25,6 +25,7 @@ async function buildGenomes(): Promise<BRCDataCatalogGenome[]> {
       chromosomes: parseNumberOrNull(row.chromosomeCount),
       coverage: parseStringOrNull(row.coverage),
       gcPercent: parseNumber(row.gcPercent),
+      geneModelUrl: parseStringOrNull(row.geneModelUrl),
       isRef: parseBoolean(row.isRef),
       length: parseNumber(row.length),
       level: row.level,

diff --git a/files/build-files-from-ncbi.py b/files/build-files-from-ncbi.py
@@ -1,6 +1,7 @@
-from urllib.parse import quote as url_quote
 import pandas as pd
 import requests
+import urllib.parse
+import re
 
 TAXA_URL = "https://docs.google.com/spreadsheets/d/1Gg9sw2Qw765tOx2To53XkTAn-RAMiBtqYrfItlLXXrc/gviz/tq?tqx=out:csv&sheet=Sheet1.csv"
 
@@ -28,7 +29,7 @@ def get_tax_ids(organisms_df):
   return list(organisms_df["taxonomyId"])
 
 def build_genomes_url(tax_ids):
-  return f"https://api.ncbi.nlm.nih.gov/datasets/v2/genome/taxon/{url_quote(",".join([str(id) for id in tax_ids]))}/dataset_report?filters.assembly_source=refseq&filters.has_annotation=true&filters.exclude_paired_reports=true&filters.exclude_atypical=true&filters.assembly_level=scaffold&filters.assembly_level=chromosome&filters.assembly_level=complete_genome"
+  return f"https://api.ncbi.nlm.nih.gov/datasets/v2/genome/taxon/{urllib.parse.quote(",".join([str(id) for id in tax_ids]))}/dataset_report?filters.assembly_source=refseq&filters.has_annotation=true&filters.exclude_paired_reports=true&filters.exclude_atypical=true&filters.assembly_level=scaffold&filters.assembly_level=chromosome&filters.assembly_level=complete_genome"
 
 def get_genome_row(genome_info):
   refseq_category = genome_info["assembly_info"].get("refseq_category")
@@ -52,6 +53,38 @@ def get_genome_row(genome_info):
 def get_genomes_df(tax_ids):
   return pd.DataFrame(data=[get_genome_row(genome_info) for genome_info in requests.get(build_genomes_url(tax_ids)).json()["reports"]])
 
+def _id_to_gene_model_url(asm_id):
+  hubs_url = "https://hgdownload.soe.ucsc.edu/hubs/"
+  components = [asm_id[0:3], asm_id[4:7], asm_id[7:10], asm_id[10:13], asm_id, "genes"]
+  url = urllib.parse.urljoin(hubs_url, "/".join(components))
+  # url looks something like https://hgdownload.soe.ucsc.edu/hubs/GCF/030/504/385/GCF_030504385.1/genes/
+  # and contains html content with links to gene models.
+  # we need to scrape this to get the gtf
+  print(f"fetching url {url}")
+  response = requests.get(url)
+  try:
+    response.raise_for_status()
+  except Exception:
+    # FIXME?: Some accessions don't have a gene folder
+    return None
+  # find link to gtf, should ideally be ncbiRefSeq, but augustus will do
+  html_content = response.text
+  pattern = rf"{asm_id.replace('.', r'\.')}.*?\.gtf\.gz"
+  augustus_file = None
+  for match in re.findall(pattern, html_content):
+    if "ncbiRefSeq" in match:
+      return urllib.parse.urljoin(f"{url}/", match)
+    elif "augustus" in match:
+      augustus_file = match
+  if augustus_file:
+    return urllib.parse.urljoin(f"{url}/", augustus_file)
+  # No match, I guess that's OK ?
+  return None
+
+
+def add_gene_model_url(genomes_df: pd.DataFrame):
+  return pd.concat([genomes_df, genomes_df["accession"].apply(_id_to_gene_model_url).rename("geneModelUrl")], axis="columns")
+
 def build_files():
   print("Building files")
 
@@ -71,7 +104,7 @@ def build_files():
   gen_bank_merge_df = genomes_source_df.merge(assemblies_df, how="left", left_on="pairedAccession", right_on="genBank")
   ref_seq_merge_df = genomes_source_df.merge(assemblies_df, how="left", left_on="accession", right_on="refSeq")
 
-  genomes_df = gen_bank_merge_df.combine_first(ref_seq_merge_df)
+  genomes_df = add_gene_model_url(gen_bank_merge_df.combine_first(ref_seq_merge_df))
 
   genomes_df.to_csv(GENOMES_OUTPUT_PATH, index=False, sep="\t")
 

diff --git a/files/entities.ts b/files/entities.ts
@@ -4,6 +4,7 @@ export interface SourceGenome {
   chromosomeCount: string;
   coverage: string;
   gcPercent: string;
+  geneModelUrl: string;
   isRef: string;
   length: string;
   level: string;

diff --git a/files/out/genomes.json b/files/out/genomes.json
@@ -5,6 +5,7 @@
     "chromosomes": 14,
     "coverage": null,
     "gcPercent": 42.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/415/GCF_000002415.2/genes/GCF_000002415.2_ASM241v2.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 27007701,
     "level": "Chromosome",
@@ -21,6 +22,7 @@
     "chromosomes": 11,
     "coverage": null,
     "gcPercent": 46.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/445/GCF_000002445.2/genes/GCF_000002445.2_ASM244v1.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 26075494,
     "level": "Chromosome",
@@ -37,6 +39,7 @@
     "chromosomes": 36,
     "coverage": null,
     "gcPercent": 59.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/725/GCF_000002725.2/genes/GCF_000002725.2_ASM272v2.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 32855089,
     "level": "Complete Genome",
@@ -53,6 +56,7 @@
     "chromosomes": 14,
     "coverage": "100.0x",
     "gcPercent": 19.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/765/GCF_000002765.6/genes/GCF_000002765.6_GCA_000002765.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 23292622,
     "level": "Complete Genome",
@@ -69,6 +73,7 @@
     "chromosomes": 35,
     "coverage": null,
     "gcPercent": 58,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/845/GCF_000002845.2/genes/GCF_000002845.2_ASM284v2.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 32068771,
     "level": "Chromosome",
@@ -85,6 +90,7 @@
     "chromosomes": 14,
     "coverage": "26.5x",
     "gcPercent": 52.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/006/565/GCF_000006565.2/genes/GCF_000006565.2_TGA4.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 65633124,
     "level": "Chromosome",
@@ -101,6 +107,7 @@
     "chromosomes": null,
     "coverage": null,
     "gcPercent": 46,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/149/335/GCF_000149335.2/genes/GCF_000149335.2_ASM14933v2.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 28947925,
     "level": "Scaffold",
@@ -117,6 +124,7 @@
     "chromosomes": 1,
     "coverage": null,
     "gcPercent": 65.5,
+    "geneModelUrl": null,
     "isRef": true,
     "length": 4411532,
     "level": "Complete Genome",
@@ -133,6 +141,7 @@
     "chromosomes": null,
     "coverage": null,
     "gcPercent": 51.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/209/065/GCF_000209065.1/genes/GCF_000209065.1_ASM20906v1.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 89937456,
     "level": "Scaffold",
@@ -149,6 +158,7 @@
     "chromosomes": 36,
     "coverage": null,
     "gcPercent": 59.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/000/227/135/GCF_000227135.1/genes/GCF_000227135.1_ASM22713v2.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 32444968,
     "level": "Chromosome",
@@ -165,6 +175,7 @@
     "chromosomes": 1,
     "coverage": null,
     "gcPercent": 65.5,
+    "geneModelUrl": null,
     "isRef": false,
     "length": 4411709,
     "level": "Complete Genome",
@@ -181,6 +192,7 @@
     "chromosomes": 1,
     "coverage": null,
     "gcPercent": 33,
+    "geneModelUrl": null,
     "isRef": false,
     "length": 196858,
     "level": "Complete Genome",
@@ -197,6 +209,7 @@
     "chromosomes": 1,
     "coverage": null,
     "gcPercent": 38,
+    "geneModelUrl": null,
     "isRef": false,
     "length": 29903,
     "level": "Complete Genome",
@@ -213,6 +226,7 @@
     "chromosomes": 3,
     "coverage": "250.0x",
     "gcPercent": 37,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/016/801/865/GCF_016801865.2/genes/GCF_016801865.2_TS_CPP_V2.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 566339288,
     "level": "Chromosome",
@@ -229,6 +243,7 @@
     "chromosomes": 9,
     "coverage": "475.0x",
     "gcPercent": 46.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/018/416/015/GCF_018416015.2/genes/GCF_018416015.2_ASM1841601v2.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 28193268,
     "level": "Complete Genome",
@@ -245,6 +260,7 @@
     "chromosomes": 1,
     "coverage": "20.0x",
     "gcPercent": 65.5,
+    "geneModelUrl": null,
     "isRef": false,
     "length": 4516435,
     "level": "Complete Genome",
@@ -261,6 +277,7 @@
     "chromosomes": 14,
     "coverage": "100.0x",
     "gcPercent": 21.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/900/002/385/GCF_900002385.2/genes/GCF_900002385.2_GCA_900002385.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 23043114,
     "level": "Complete Genome",
@@ -277,6 +294,7 @@
     "chromosomes": 14,
     "coverage": "155.0x",
     "gcPercent": 23,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/900/681/995/GCF_900681995.1/genes/GCF_900681995.1_PVVCY_v1.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 18338688,
     "level": "Complete Genome",
@@ -293,6 +311,7 @@
     "chromosomes": 3,
     "coverage": "54.0x",
     "gcPercent": 44.5,
+    "geneModelUrl": "https://hgdownload.soe.ucsc.edu/hubs/GCF/943/734/735/GCF_943734735.2/genes/GCF_943734735.2_idAnoGambNW_F1_1.ncbiRefSeq.gtf.gz",
     "isRef": true,
     "length": 264451381,
     "level": "Chromosome",
@@ -309,6 +328,7 @@
     "chromosomes": 1,
     "coverage": "100.0x",
     "gcPercent": 65.5,
+    "geneModelUrl": null,
     "isRef": false,
     "length": 4469156,
     "level": "Complete Genome",

diff --git a/files/source/genomes-from-ncbi.tsv b/files/source/genomes-from-ncbi.tsv
@@ -1,21 +1,21 @@
-taxon	taxonomyId	accession	isRef	level	chromosomeCount	length	scaffoldCount	scaffoldN50	scaffoldL50	coverage	gcPercent	annotationStatus	pairedAccession	ucscBrowser	genBank	refSeq
-Mycobacterium tuberculosis H37Rv	83332	GCF_000195955.2	True	Complete Genome	1.0	4411532	1	4411532	1		65.5		GCA_000195955.2			
-Plasmodium falciparum 3D7	36329	GCF_000002765.6	True	Complete Genome	14.0	23292622	14	1687656	5	100.0x	19.5	Full annotation	GCA_000002765.3	https://genome.ucsc.edu/h/GCF_000002765.5	GCA_000002765.3	GCF_000002765.5
-Leishmania major strain Friedlin	347515	GCF_000002725.2	True	Complete Genome	36.0	32855089	36	1091540	11		59.5	Full annotation	GCA_000002725.2	https://genome.ucsc.edu/h/GCF_000002725.2	GCA_000002725.2	GCF_000002725.2
-Plasmodium yoelii	5861	GCF_900002385.2	True	Complete Genome	14.0	23043114	14	2046250	5	100.0x	21.5	Full annotation	GCA_900002385.2	https://genome.ucsc.edu/h/GCF_900002385.2	GCA_900002385.2	GCF_900002385.2
-Coccidioides posadasii str. Silveira	443226	GCF_018416015.2	True	Complete Genome	9.0	28193268	9	8079863	2	475.0x	46.5	Full annotation	GCA_018416015.2	https://genome.ucsc.edu/h/GCA_018416015.2	GCA_018416015.2	GCF_018416015.1
-Plasmodium vinckei vinckei	54757	GCF_900681995.1	True	Complete Genome	14.0	18338688	14	1692345	5	155.0x	23.0	Full annotation	GCA_900681995.1	https://genome.ucsc.edu/h/GCF_900681995.1	GCA_900681995.1	GCF_900681995.1
-Leishmania donovani	5661	GCF_000227135.1	True	Chromosome	36.0	32444968	36	1024085	11		59.5	Full annotation	GCA_000227135.2	https://genome.ucsc.edu/h/GCF_000227135.1	GCA_000227135.2	GCF_000227135.1
-Toxoplasma gondii ME49	508771	GCF_000006565.2	True	Chromosome	14.0	65633124	2276	4973582	6	26.5x	52.5		GCA_000006565.2	https://genome.ucsc.edu/h/GCF_000006565.2	GCA_000006565.2	GCF_000006565.2
-Trypanosoma brucei brucei TREU927	185431	GCF_000002445.2	True	Chromosome	11.0	26075494	12	2481190	4		46.5	Full annotation	GCA_000002445.1	https://genome.ucsc.edu/h/GCF_000002445.2	GCA_000002445.1	GCF_000002445.2
-Anopheles gambiae	7165	GCF_943734735.2	True	Chromosome	3.0	264451381	190	99149756	2	54.0x	44.5	Full annotation	GCA_943734735.2	https://genome.ucsc.edu/h/GCF_943734735.2	GCA_943734735.2	GCF_943734735.2
-Plasmodium vivax	5855	GCF_000002415.2	True	Chromosome	14.0	27007701	2747	1678596	6		42.5	Full annotation	GCA_000002415.2	https://genome.ucsc.edu/h/GCF_000002415.2	GCA_000002415.2	GCF_000002415.2
-Culex pipiens pallens	42434	GCF_016801865.2	True	Chromosome	3.0	566339288	289	186194774	2	250.0x	37.0	Full annotation	GCA_016801865.2			
-Leishmania braziliensis MHOM/BR/75/M2904	420245	GCF_000002845.2	True	Chromosome	35.0	32068771	138	992961	11		58.0		GCA_000002845.2	https://genome.ucsc.edu/h/GCF_000002845.2	GCA_000002845.2	GCF_000002845.2
-Trypanosoma cruzi	5693	GCF_000209065.1	True	Scaffold		89937456	29495	88624	212		51.5	Full annotation	GCA_000209065.1	https://genome.ucsc.edu/h/GCF_000209065.1	GCA_000209065.1	GCF_000209065.1
-Coccidioides immitis RS	246410	GCF_000149335.2	True	Scaffold		28947925	6	4323945	3		46.0	Full annotation	GCA_000149335.2	https://genome.ucsc.edu/h/GCF_000149335.2	GCA_000149335.2	GCF_000149335.2
-Mycobacterium tuberculosis H37Rv	83332	GCF_000277735.2	False	Complete Genome	1.0	4411709	1	4411709	1		65.5		GCA_000277735.2			
-Severe acute respiratory syndrome coronavirus 2	2697049	GCF_009858895.2	False	Complete Genome	1.0	29903	1	29903	1		38.0		GCA_009858895.3			
-Monkeypox virus	10244	GCF_000857045.1	False	Complete Genome	1.0	196858	1	196858	1		33.0		GCA_000857045.1			
-Mycobacterium tuberculosis	1773	GCF_030566675.1	False	Complete Genome	1.0	4516435	1	4516435	1	20.0x	65.5		GCA_030566675.1			
-Mycobacterium tuberculosis	1773	GCF_963525475.1	False	Complete Genome	1.0	4469156	1	4469156	1	100.0x	65.5		GCA_963525475.1			
+taxon	taxonomyId	accession	isRef	level	chromosomeCount	length	scaffoldCount	scaffoldN50	scaffoldL50	coverage	gcPercent	annotationStatus	pairedAccession	ucscBrowser	genBank	refSeq	geneModelUrl
+Mycobacterium tuberculosis H37Rv	83332	GCF_000195955.2	True	Complete Genome	1.0	4411532	1	4411532	1		65.5		GCA_000195955.2				
+Plasmodium falciparum 3D7	36329	GCF_000002765.6	True	Complete Genome	14.0	23292622	14	1687656	5	100.0x	19.5	Full annotation	GCA_000002765.3	https://genome.ucsc.edu/h/GCF_000002765.5	GCA_000002765.3	GCF_000002765.5	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/765/GCF_000002765.6/genes/GCF_000002765.6_GCA_000002765.ncbiRefSeq.gtf.gz
+Leishmania major strain Friedlin	347515	GCF_000002725.2	True	Complete Genome	36.0	32855089	36	1091540	11		59.5	Full annotation	GCA_000002725.2	https://genome.ucsc.edu/h/GCF_000002725.2	GCA_000002725.2	GCF_000002725.2	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/725/GCF_000002725.2/genes/GCF_000002725.2_ASM272v2.ncbiRefSeq.gtf.gz
+Plasmodium yoelii	5861	GCF_900002385.2	True	Complete Genome	14.0	23043114	14	2046250	5	100.0x	21.5	Full annotation	GCA_900002385.2	https://genome.ucsc.edu/h/GCF_900002385.2	GCA_900002385.2	GCF_900002385.2	https://hgdownload.soe.ucsc.edu/hubs/GCF/900/002/385/GCF_900002385.2/genes/GCF_900002385.2_GCA_900002385.ncbiRefSeq.gtf.gz
+Coccidioides posadasii str. Silveira	443226	GCF_018416015.2	True	Complete Genome	9.0	28193268	9	8079863	2	475.0x	46.5	Full annotation	GCA_018416015.2	https://genome.ucsc.edu/h/GCA_018416015.2	GCA_018416015.2	GCF_018416015.1	https://hgdownload.soe.ucsc.edu/hubs/GCF/018/416/015/GCF_018416015.2/genes/GCF_018416015.2_ASM1841601v2.ncbiRefSeq.gtf.gz
+Plasmodium vinckei vinckei	54757	GCF_900681995.1	True	Complete Genome	14.0	18338688	14	1692345	5	155.0x	23.0	Full annotation	GCA_900681995.1	https://genome.ucsc.edu/h/GCF_900681995.1	GCA_900681995.1	GCF_900681995.1	https://hgdownload.soe.ucsc.edu/hubs/GCF/900/681/995/GCF_900681995.1/genes/GCF_900681995.1_PVVCY_v1.ncbiRefSeq.gtf.gz
+Leishmania donovani	5661	GCF_000227135.1	True	Chromosome	36.0	32444968	36	1024085	11		59.5	Full annotation	GCA_000227135.2	https://genome.ucsc.edu/h/GCF_000227135.1	GCA_000227135.2	GCF_000227135.1	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/227/135/GCF_000227135.1/genes/GCF_000227135.1_ASM22713v2.ncbiRefSeq.gtf.gz
+Toxoplasma gondii ME49	508771	GCF_000006565.2	True	Chromosome	14.0	65633124	2276	4973582	6	26.5x	52.5		GCA_000006565.2	https://genome.ucsc.edu/h/GCF_000006565.2	GCA_000006565.2	GCF_000006565.2	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/006/565/GCF_000006565.2/genes/GCF_000006565.2_TGA4.ncbiRefSeq.gtf.gz
+Trypanosoma brucei brucei TREU927	185431	GCF_000002445.2	True	Chromosome	11.0	26075494	12	2481190	4		46.5	Full annotation	GCA_000002445.1	https://genome.ucsc.edu/h/GCF_000002445.2	GCA_000002445.1	GCF_000002445.2	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/445/GCF_000002445.2/genes/GCF_000002445.2_ASM244v1.ncbiRefSeq.gtf.gz
+Anopheles gambiae	7165	GCF_943734735.2	True	Chromosome	3.0	264451381	190	99149756	2	54.0x	44.5	Full annotation	GCA_943734735.2	https://genome.ucsc.edu/h/GCF_943734735.2	GCA_943734735.2	GCF_943734735.2	https://hgdownload.soe.ucsc.edu/hubs/GCF/943/734/735/GCF_943734735.2/genes/GCF_943734735.2_idAnoGambNW_F1_1.ncbiRefSeq.gtf.gz
+Plasmodium vivax	5855	GCF_000002415.2	True	Chromosome	14.0	27007701	2747	1678596	6		42.5	Full annotation	GCA_000002415.2	https://genome.ucsc.edu/h/GCF_000002415.2	GCA_000002415.2	GCF_000002415.2	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/415/GCF_000002415.2/genes/GCF_000002415.2_ASM241v2.ncbiRefSeq.gtf.gz
+Culex pipiens pallens	42434	GCF_016801865.2	True	Chromosome	3.0	566339288	289	186194774	2	250.0x	37.0	Full annotation	GCA_016801865.2				https://hgdownload.soe.ucsc.edu/hubs/GCF/016/801/865/GCF_016801865.2/genes/GCF_016801865.2_TS_CPP_V2.ncbiRefSeq.gtf.gz
+Leishmania braziliensis MHOM/BR/75/M2904	420245	GCF_000002845.2	True	Chromosome	35.0	32068771	138	992961	11		58.0		GCA_000002845.2	https://genome.ucsc.edu/h/GCF_000002845.2	GCA_000002845.2	GCF_000002845.2	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/002/845/GCF_000002845.2/genes/GCF_000002845.2_ASM284v2.ncbiRefSeq.gtf.gz
+Trypanosoma cruzi	5693	GCF_000209065.1	True	Scaffold		89937456	29495	88624	212		51.5	Full annotation	GCA_000209065.1	https://genome.ucsc.edu/h/GCF_000209065.1	GCA_000209065.1	GCF_000209065.1	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/209/065/GCF_000209065.1/genes/GCF_000209065.1_ASM20906v1.ncbiRefSeq.gtf.gz
+Coccidioides immitis RS	246410	GCF_000149335.2	True	Scaffold		28947925	6	4323945	3		46.0	Full annotation	GCA_000149335.2	https://genome.ucsc.edu/h/GCF_000149335.2	GCA_000149335.2	GCF_000149335.2	https://hgdownload.soe.ucsc.edu/hubs/GCF/000/149/335/GCF_000149335.2/genes/GCF_000149335.2_ASM14933v2.ncbiRefSeq.gtf.gz
+Mycobacterium tuberculosis H37Rv	83332	GCF_000277735.2	False	Complete Genome	1.0	4411709	1	4411709	1		65.5		GCA_000277735.2				
+Severe acute respiratory syndrome coronavirus 2	2697049	GCF_009858895.2	False	Complete Genome	1.0	29903	1	29903	1		38.0		GCA_009858895.3				
+Monkeypox virus	10244	GCF_000857045.1	False	Complete Genome	1.0	196858	1	196858	1		33.0		GCA_000857045.1				
+Mycobacterium tuberculosis	1773	GCF_030566675.1	False	Complete Genome	1.0	4516435	1	4516435	1	20.0x	65.5		GCA_030566675.1				
+Mycobacterium tuberculosis	1773	GCF_963525475.1	False	Complete Genome	1.0	4469156	1	4469156	1	100.0x	65.5		GCA_963525475.1