Trove refactor commit

.gitignore

@@ -1,3 +1,15 @@
+# OS generated files
+.DS_Store
+.DS_Store?
+._*
+.Spotlight-V100
+.Trashes
+ehthumbs.db
+Thumbs.d
+
+# Trove specific cache
+.trove/
+
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]

README.md

@@ -0,0 +1,39 @@
+# Trove 
+
+**Preprint**: [Trove: Ontology-driven Weak Supervision for Medical Entity Classification] (https://arxiv.org/abs/2008.01972)
+
+See the `manuscript` branch for the preprint's code
+
+
+Trove is a weakly supervised framework for training medical named entity recognition (NER) classifiers without hand-labeled training data.
+***Trove is currently in-development software!***. Let us know when you find bugs.
+
+## Installation
+
+Requirements: python 3.6, pytorch 1.0+, snorkel 0.9.5+
+
+## Tutorials
+
+See `tutorials/`
+
+## Requirements
+
+Tested on OSX and Linux.
+
+## Citation
+If use Trove in your research, please cite.
+
+```bibtex
+@ARTICLE{Fries2020-wg,
+   title    = "Trove: Ontology-driven weak supervision 
+               for medical entity classification",
+   author   = "Fries, Jason A and Steinberg, Ethan and Khattar, Saelig and
+               Fleming, Scott L and Posada, Jose and Callahan, Alison and Shah,
+               Nigam H",
+   journal  = "ArXiv",
+   month    =  aug,
+   year     =  2020,
+   language = "en"
+}
+```
+

applications/README.md

@@ -0,0 +1,11 @@
+# Contributed Applications
+
+###NOTE: These are currently being refactored.
+
+Labeling functions for various weakly supervised biomedical classification tasks
+
+- `bc5cdr/` - BioCreative V Chemical Disease Relations Chemical and Disease NER (literature)
+- `i2b2drugs/` - n2c2 (formally i2b2) Drug NER (clinical)
+- `shareclef2014/` - ShARe/CLEF 2014 Disorder NER (clinical)
+- `thyme/` - DocTimeRel (clinical)
+- `covid19/` - COVID-19 exposure (clinical)

applications/bc5cdr/cdr_chemical_regexes.tsv

@@ -0,0 +1,40 @@
+LABEL	NAME	TIER	IGNORE_CASE	REGEX	NOTES
+0	non_chemicals	4	1	\b([A-Za-z0-9]+?[rlntd]ase[s]*)\b
+0	non_chemicals	4	1	[A-Za-z0-9]+ factor[s]*
+0	non_chemicals	4	1	(angiotensinogen)
+0	non_chemicals	4	1	(guarana|(panax )*ginseng)
+0	non_chemicals	4	1	\b(anti[a-z]+)\b
+0	non_chemicals	4	1	(cocaine (abuse|addiction|overdose))
+0	non_chemicals	4	1	((renal)\s*(angiotensinogen) (mRNA|expression))
+0	non_chemicals	4	1	(atrial natriuretic factor( \[\s*ANF\s*\])*)
+0	non_chemicals	4	1	(fibrinolysis inhibitor[s]*)
+0	non_chemicals	4	1	((brain )*biogenic amines)
+0	non_chemicals	4	1	([-]\s*(associated|dependent|related|treated|acting|controlled|induced|containing|fold|increasing|adjusted|month|specific))
+1	misc_chemicals	4	1	(oral contraceptives)
+1	misc_chemicals	4	1	([A-Z]){2}[0-9]{3,}
+1	misc_chemicals	4	1	\b(ACEi|ACE inhibitor[s]*)\b
+1	misc_chemicals	4	1	(corticosteroid[s]*|(oral[- ])*contraceptive[s]*)
+1	misc_chemicals	4	1	(calcium|cacl[(]2[)])
+1	misc_chemicals	4	1	([l][- ](glutathione|arginine))
+1	misc_chemicals	4	1	(appetite[- ]suppressant[s]*( drugs)*)
+1	misc_chemicals	4	1	(calcium channel blocker[s]*|calcium chloride|CaCl)
+1	misc_chemicals	4	1	(simvastatin[- ]ezetimibe)
+1	misc_chemicals	4	1	([snp][-](perillyl alcohol|pyrimidinyl|choloroaniline|acetylcysteine|limonene))
+1	misc_chemicals	4	1	((alpha|beta|gamma)[-][T])
+1	misc_chemicals	4	0	(PG[-]9|U[-]II)
+1	misc_chemicals	4	0	(BPO|GSH|DFU|CsA|Srl|HOE|GVG|PAN|NMDA)
+1	misc_chemicals	4	0	(TCR|MZ|HBsAg|AraG|LR132|SSRI[s]*|HBeAg|LR132|BD10[0-9]{2}|GNC92H2|SSR103800|CGRP)
+1	misc_chemicals	4	1	(angiotensin([- ]ii)*)
+1	misc_chemicals	4	1	(u[- ]ii|urotensin[- ]ii)
+1	misc_chemicals	4	1	bradykinin	comprised of 9 amino acids
+1	misc_chemicals	4	1	(d[- ]pen(icillamine)*)
+1	misc_chemicals	4	1	(lipopolysaccharide|alkylating agents)
+1	misc_chemicals	4	1	(pegylated (interferon|IFN)( alpha[- ]2[ab])*)
+1	misc_chemicals	4	1	(\[3H\])
+1	misc_chemicals	4	1	(CaCl|LAM|GSH|PAN|H2O|AVP|LR132)
+1	glue_tokens	4	1	(\[[0-9][-.])
+1	glue_tokens	4	1	(\[\s*3H\s*\])
+1	glue_tokens	4	1	(thiazolyl|amino|phenyl|ethyl|butyl|nonane|3H)[\]]
+1	glue_tokens	4	1	([-](methyl|ethyl|carboline|dimethoxy|alpha|beta|delta|gamma|glyceryl|thiazolyl)[-])
+1	glue_tokens	4	0	[-]([1-9]|[A-Z])[-]
+0	parentheses	4	1	[(](P|p|n)\s*([><=]+|(less|great)(er)*)|(ml|mg|kg|g|(year|day|month)[s]*)[)]|[(][0-9]+[%][)]

applications/bc5cdr/cdr_disease_regexes.tsv

@@ -0,0 +1,31 @@
+LABEL	NAME	TIER	IGNORE_CASE	REGEX	NOTES
+1	char_grams	4	1	^(psych|necro|nephro|hyper|throm|hypo|acro|hemo)[A-Za-z]+?([rlt]ic)$
+1	char_grams	4	1	^(hepato|hemato|nephro|cardio|neuro|myelo|oto)*toxic(ities|ity)*$
+1	diseases_1	4	1	(([A-Za-z]+\s*) and ([A-Za-z]+\s*) ((neuro)*toxicity|injury|lesion[s]*|impairment|effusion[s]*|deficit[s]*))
+1	diseases_1	4	1	([A-Za-z]+) and ([A-Za-z]+) (insufficiency|(dysfunction|carcinoma|cancer|syndrome|disorder|disease)[s]*)
+1	diseases_1	4	1	\b(non[-](small|hodgkin)|veno[-]occlusive|end[-]stage|HBV[-]HIV|Q[-]T)\b	# hyphens
+1	diseases_1	4	1	(increase[s]* in (blood pressure|heart rate|locomotor activity|dural( and cortical) blood flow))	# increases/decreases in X
+1	diseases_1	4	1	((reduction|decrease)[s*] in (MAP|glomerular number|(arterial )*blood pressure))	# increases/decreases in X
+1	diseases_1	4	1	((respiratory|hypothalamic|corticostriatal|tubular|biventricular|myocardial|hepatic|systolic|cranial nerve|sexual) dysfunction[s]*)	# increases/decreases in X
+1	diseases_1	4	1	(myocardial( cell)*|hepatocellular|mitochondrial|proteinuric|hippocampal|cerebellum|myocardial|neuronal|cardiac|hepatic|bladder|tissue|axonal|kidney|renal|liver|cord) (injury|damage)	# injuries
+1	diseases_1	4	1	(malignant ([A-Za-z]+ )*(glioma|tumor)[s]*)
+1	diseases_1	4	1	(([A-Za-z]+)'s|wolff[- ]+parkinson[- ]+white|haemolytic[- ]+uraemic|guillain[- ]+barr|hematologic|cholestatic|rabbit)([- ]+like)* syndrome
+1	diseases_1	4	1	diabetic( hyperalgesia)*|diabetes
+1	diseases_1	4	1	(adenocarcinoma|calcification|angiosarcoma|enlargement|disorders|cirrhosis|carcinoma|cancer|injury) (in|of) the (central nervous system|oral cavity|bladder|artery|ureter|brain|aorta|liver)	# anatomy findings
+1	diseases_1	4	1	\b[A-Za-z-]+'s (syndrome|disease)\b	# common disease patterns
+1	diseases_1	4	1	((artery )*calcification)|(calcification of the [A-Za-z]+)	# common disease patterns
+1	diseases_1	4	1	([Dd]uchenne('s)* (muscular )*dystrophy|DMD)	# dystrophy
+1	diseases_1	4	1	(ventricular tachyarrhythmias|loss of consciousness|tachyarrhythmias|hyperhidrosis|hypertensive|cardiomegaly|weight gain|hypotension|weight loss|glucosuria|hoarding)	# common findings
+1	diseases_2	4	1	(hyperactive|convulsive|haemorrhage|depressed|deformation[s]*)
+1	diseases_2	4	1	\b((sugar|drug) dependency|nicotine-induced nystagmus|nystagmus|NIN)\b
+1	diseases_2	4	1	(weakness of extremities|transverse limb deficiency|increase in locomotor activity|palpebral twitching)	# movement/muscule issues
+1	diseases_2	4	1	(choreoathetoid movement[s]*|choreatiform hyperkinesias)	# movement/muscule issues
+1	diseases_2	4	1	(tender joints|tenderness|swelling|morning stiffness|excessive flexion)	# movement/muscule issues             
+1	diseases_2	4	1	(valve|valvular|valvular heart) (regurgitation|abnormalit(y|ies))	# cardiac
+1	diseases_2	4	1	(atherosclerotic obstruction|cardiac remodelling)	# cardiac           
+1	diseases_2	4	1	(cholestatic|renovascular|renal and kidney) disease[s]*	# neurological/renal
+1	diseases_2	4	1	(cranial nerve|hepatic and renal|cardiac|renal) dysfunction[s]*	# neurological/renal
+1	diseases_2	4	1	(neuronal loss|cranial nerve deficits|hippocampal injury|behavioral abnormalities|deficits in communication|repetitive behaviors|impaired immediate free recall)	# neurological/renal
+1	diseases_2	4	1	(vanishing bile duct|renal and hepatic failure|hepatic impairment|deterioration of renal function|abnormal liver function)	# neurological/renal
+0	non_diseases	4	1	([-]\s*(associated|dependent|related|treated|acting|controlled|induced|containing|fold|increasing|adjusted|month|specific))	# drug induced / associated effects aren't labeled
+0	non_diseases	4	1	(toxic ((side )*effect[s]*|agent[s]*|action|state|reaction|range|death[s]*|profile|assault[s]*)|(highly|minimally) toxic)	# toxic effects aren't diseases