LRT Method Application: Part 1 / Zhihao Yao

Inputs.table(dataMatch)

url = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+patient.drug.openfda.generic_name:(%22'+drugName+'%22)+AND+receivedate:(['+dateFrom+'+TO+'+dateTo+'])&limit=1000&count=patient.reaction.reactionmeddrapt.exact&skip=0'

raw = await fetch(url).then(res => res.json())

rawTable = Inputs.table(raw.results)

url2 = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+_exists_:(patient.reaction.reactionmeddrapt.exact)+AND+_exists_:(patient.drug.openfda.generic_name)+AND+receivedate:(['+dateFrom+'+TO+'+dateTo+'])&count=patient.reaction.reactionmeddrapt.exact&limit=1000&sort=patient.reaction.reactionmeddrapt.exact:desc'

raw2 = await fetch(url2).then(res => res.json())

raw2Table = Inputs.table(raw2.results)

dataTop = raw.results.sort((a,b) => b.count - a.count)

.filter((n,i) => i < maxNumberOfEvent)

.map(n => ({

AE: n.term,

ntrt: +n.count

}));

dataTopTerm = [...new Set(dataTop.map(n => n.AE))]

dataOther = raw.results.sort((a,b) => b.count - a.count)

.filter((n,i) => i >= maxNumberOfEvent)

.map(n => ({

AE: n.term,

ntrt: +n.count

}));

countDrugOtherTotal = dataOther.reduce((p,c)=>{p += c.ntrt; return p;}, 0);

dataAll = raw2.results.sort((a,b) => b.count - a.count)

.map(n => ({

AE: n.term,

nidot: +n.count

}));

dataUnMatch.view()

dataUnMatch = aq.from(dataTop).antijoin(aq.from(dataAll), ['AE', 'AE'], [aq.all(), aq.not('AE')]);

dataUnMatchTerm = [...new Set(dataUnMatch.objects().map(n => n.AE))];

drugAllFinal = {

for (let k of dataUnMatchTerm) {

var kStr = k.split(' ').join('%20');

var urlK = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+_exists_:(patient.drug.openfda.generic_name)+AND+patient.reaction.reactionmeddrapt.exact:(%22'+kStr+'%22)+AND+receivedate:(['+dateFrom+'+TO+'+dateTo+'])&limit=1&skip=0';

var dataK = await fetch(urlK).then(res => res.json())

dataAll.push({'AE': k, 'nidot': dataK.meta.results.total});

}

return dataAll;

}

countDrugAllFinalOther = drugAllFinal.filter(n => dataTopTerm.indexOf(n.AE) == -1).reduce((p,c) => {p += c.nidot; return p;},0);

dataMatch = {

var dataMatch = aq.from(dataTop).join_left(aq.from(drugAllFinal), ['AE', 'AE'], [aq.all(), aq.not('AE')]).objects()

.filter(n => n.nidot !== undefined);

dataMatch.push({

'AE': 'Other',

'ntrt': countDrugOtherTotal,

'nidot': countDrugAllFinalOther

});

dataMatch.map(n => n.ndotj = dataMatch.reduce((p,c) => {p += c.ntrt; return p;}, 0));

dataMatch.map(n => n.ndotdot = dataMatch.reduce((p,c) => {p += c.nidot; return p;}, 0));

return dataMatch;

}

LRTDataTable = Inputs.table(LRTData)

async function generateLRTData() {

var url = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+patient.drug.openfda.generic_name:(%22'+drugName+'%22)+AND+receivedate:(['+dateFrom+'+TO+'+dateTo+'])&limit=1000&count=patient.reaction.reactionmeddrapt.exact&skip=0';

var raw = await fetch(url).then(res => res.json());

var url2 = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+_exists_:(patient.reaction.reactionmeddrapt.exact)+AND+_exists_:(patient.drug.openfda.generic_name)+AND+receivedate:(['+dateFrom+'+TO+'+dateTo+'])&count=patient.reaction.reactionmeddrapt.exact&limit=1000&sort=patient.reaction.reactionmeddrapt.exact:desc';

var raw2 = await fetch(url2).then(res => res.json());

var dataTop = raw.results.sort((a,b) => b.count - a.count).filter((n,i) => i < maxNumberOfEvent).map(n => ({AE: n.term,ntrt: +n.count}));

var dataTopTerm = [...new Set(dataTop.map(n => n.AE))];

var dataOther = raw.results.sort((a,b) => b.count - a.count).filter((n,i) => i >= maxNumberOfEvent).map(n => ({AE: n.term,ntrt: +n.count}));

var countDrugOtherTotal = dataOther.reduce((p,c)=>{p += c.ntrt; return p;}, 0);

var dataAll = raw2.results.sort((a,b) => b.count - a.count).map(n => ({AE: n.term,nidot: +n.count}));

var dataUnMatch = aq.from(dataTop).antijoin(aq.from(dataAll), ['AE', 'AE'], [aq.all(), aq.not('AE')]);

var dataUnMatchTerm = [...new Set(dataUnMatch.objects().map(n => n.AE))];

for (let k of dataUnMatchTerm) {

var kStr = k.split(' ').join('%20');

var urlK = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+_exists_:(patient.drug.openfda.generic_name)+AND+patient.reaction.reactionmeddrapt.exact:(%22'+kStr+'%22)+AND+receivedate:(['+dateFrom+'+TO+'+dateTo+'])&limit=1&skip=0';

var dataK = await fetch(urlK).then(res => res.json())

dataAll.push({'AE': k, 'nidot': dataK.meta.results.total});

}

var countDrugAllFinalOther = dataAll.filter(n => dataTopTerm.indexOf(n.AE) == -1).reduce((p,c) => {p += c.nidot; return p;},0);

var dataMatch = aq.from(dataTop).join_left(aq.from(dataAll), ['AE', 'AE'], [aq.all(), aq.not('AE')]).objects()

.filter(n => n.nidot !== undefined);

dataMatch.push({

'AE': 'Other',

'ntrt': countDrugOtherTotal,

'nidot': countDrugAllFinalOther

});

dataMatch.map(n => n.ndotj = dataMatch.reduce((p,c) => {p += c.ntrt; return p;}, 0));

dataMatch.map(n => n.ndotdot = dataMatch.reduce((p,c) => {p += c.nidot; return p;}, 0));

return dataMatch;

}

async function generateLRTYearData() {

var LRTYearDataContainer = [];

for(var i of yearList){

var url = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+patient.drug.openfda.generic_name:(%22'+drugName+'%22)+AND+receivedate:(['+i+'-01-01+TO+'+i+'-12-31])&limit=1000&count=patient.reaction.reactionmeddrapt.exact&skip=0';

var raw = await fetch(url).then(res => res.json());

var url2 = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+_exists_:(patient.reaction.reactionmeddrapt.exact)+AND+_exists_:(patient.drug.openfda.generic_name)+AND+receivedate:(['+i+'-01-01+TO+'+i+'-12-31])&count=patient.reaction.reactionmeddrapt.exact&limit=1000&sort=patient.reaction.reactionmeddrapt.exact:desc';

var raw2 = await fetch(url2).then(res => res.json());

var dataTop = raw.results.sort((a,b) => b.count - a.count).filter((n,i) => i < maxNumberOfEvent).map(n => ({Year: i.toString(), AE: n.term,ntrt: +n.count}));

var dataTopTerm = [...new Set(dataTop.map(n => n.AE))];

var dataOther = raw.results.sort((a,b) => b.count - a.count).filter((n,i) => i >= maxNumberOfEvent).map(n => ({Year: i.toString(),AE: n.term,ntrt: +n.count}));

var countDrugOtherTotal = dataOther.reduce((p,c)=>{p += c.ntrt; return p;}, 0);

var dataAll = raw2.results.sort((a,b) => b.count - a.count).map(n => ({Year: i, AE: n.term, nidot: +n.count}));

var dataUnMatch = aq.from(dataTop).antijoin(aq.from(dataAll), ['AE', 'AE'], [aq.all(), aq.not('AE')]);

var dataUnMatchTerm = [...new Set(dataUnMatch.objects().map(n => n.AE))];

for (var k of dataUnMatchTerm) {

var kStr = k.split(' ').join('%20');

var urlK = 'https://api.fda.gov/drug/event.json?search=_exists_:(patient.drug.openfda.generic_name)+AND+_exists_:(patient.drug.openfda.generic_name)+AND+patient.reaction.reactionmeddrapt.exact:(%22'+kStr+'%22)+AND+receivedate:(['+i+'-01-01+TO+'+i+'-12-31])&limit=1&skip=0';

var dataK = await fetch(urlK).then(res => res.json())

dataAll.push({'AE': k, 'nidot': dataK.meta.results.total});

}

var countDrugAllFinalOther = dataAll.filter(n => dataTopTerm.indexOf(n.AE) == -1).reduce((p,c) => {p += c.nidot; return p;},0);

var dataMatch = aq.from(dataTop).join_left(aq.from(dataAll), ['AE', 'AE'], [aq.all(), aq.not(['Year', 'AE'])]).objects()

.filter(n => n.nidot !== undefined);

dataMatch.push({

'Year': i.toString(),

'AE': 'Other',

'ntrt': countDrugOtherTotal,

'nidot': countDrugAllFinalOther

});

dataMatch.map(n => n.ndotj = dataMatch.reduce((p,c) => {p += c.ntrt; return p;}, 0));

dataMatch.map(n => n.ndotdot = dataMatch.reduce((p,c) => {p += c.nidot; return p;}, 0));

LRTYearDataContainer.push(...dataMatch);

}

return LRTYearDataContainer;

}

import { aq, op } from '@uwdata/arquero'

// import { py, pyodide } from '@gnestor/pyodide'

// funcLRT = py`import numpy as np

// import pandas as pd

// def flogLRv(x, fnidot, fndotj, fndotdot):

// logLRvr = x * (np.log(x) - np.log(fnidot))

// logLRvr = logLRvr + (fndotj - x) * (np.log(fndotj - x) - np.log(fndotdot - fnidot))

// logLRvr = logLRvr - fndotj * (np.log(fndotj) - np.log(fndotdot));

// logLRvr.fillna(0)

// pj = x / fnidot

// qj = (fndotj - x) / (fndotdot - fnidot)

// logLRvr[ pj < qj ] = 0

// return logLRvr

// def fLRT(qnum, number, alpha = 0.05, alphaSpending = 1):

// mylist = qnum

// mylist["ntrt"] = pd.to_numeric(mylist["ntrt"])

// mylist["nidot"] = pd.to_numeric(mylist["nidot"])

// mylist["ndotj"] = pd.to_numeric(mylist["ndotj"])

// mylist["ndotdot"] = pd.to_numeric(mylist["ndotdot"])

// mylist["logLR"] = flogLRv(mylist["ntrt"], mylist["nidot"], mylist["ndotj"][0], mylist["ndotdot"][0])

// mylist = mylist.sort_values("logLR", ascending=False)

// cnstR = number

// np.random.seed(123)

// mylist["Pvector"] = mylist["nidot"] / mylist["ndotdot"]

// sim_cntdata = np.random.multinomial(mylist["ndotj"][0], mylist["Pvector"], size = cnstR)

// logLRsim = np.array([flogLRv(i, mylist["nidot"], mylist["ndotj"][0], mylist["ndotdot"][0]) for i in sim_cntdata])

// maxlogLRsim = np.array([np.max(i) for i in logLRsim])

// mylist["T_alpha"] = np.quantile(maxlogLRsim, 1 - alpha)

// mylist["pvalue"] = np.array([ np.mean(maxlogLRsim > i) for i in mylist["logLR"]])

// if alphaSpending >= 1:

// mylist["is_signal"] = (mylist["logLR"] > mylist["T_alpha"])

// else:

// mylist["alpha"] = alphaSpending

// mylist["is_signal"] = (mylist["pvalue"] < alphaSpending)

// mylist = mylist.sort_values("logLR", ascending=False)

// return mylist

// fLRT`

// LRTResult = py`import pandas as pd

// ${funcLRT}(pd.read_json(${JSON.stringify(LRTData)}),100).to_json(orient="records")`

// LRTResultJSON = JSON.parse(LRTResult)

LRTResultJSONImport = await FileAttachment("LRTResultJSON.json").json()

LRTResultTable = Inputs.table(

LRTResultJSONImport,

// JSON.parse(LRTResult),

{columns: ["AE", "ntrt", "nidot", "ndotj", "ndotdot", "logLR", "T_alpha", "pvalue", "is_signal"],

width: {AE: 250},

// height: 550,

align: {

ntrt: "center",

nidot: "center",

ndotj: "center",

ndotdot: "center",

logLR: "center",

T_alpha: "center",

pvalue: "center",

is_signal: "center"

}

})

// LRTResultYear = py`import pandas as pd

// import numpy as np

// LRT_result_Year = pd.DataFrame()

// for k in range(len(${yearList})):

// LRTYearData = pd.read_json(${JSON.stringify(LRTYearData)})

// countAeListYeark = LRTYearData[LRTYearData["Year"] <= ${yearList}[k]].groupby("AE").agg({"ntrt": "sum"})

// countAeListTotalYeark = LRTYearData[LRTYearData["Year"] <= ${yearList}[k]].groupby("AE").agg({"nidot": "sum"})

// combine_data_Yeark = pd.concat([countAeListYeark, countAeListTotalYeark], axis=1, join="inner")

// combine_data_Yeark["ndotj"] = np.sum(combine_data_Yeark["ntrt"])

// combine_data_Yeark["ndotdot"] = np.sum(combine_data_Yeark["nidot"])

// alphaSpending = 0.05 / (2**k)

// LRT_result_Yeark = ${funcLRT}(combine_data_Yeark, 100, alphaSpending = alphaSpending)

// LRT_result_Yeark["year"] = str(${yearList}[k])

// LRT_result_Year = LRT_result_Year.append(LRT_result_Yeark)

// LRT_result_Year['AE'] = LRT_result_Year.index

// LRT_result_Year.to_json(orient="records")`

// LRTResultYearJSON = JSON.parse(LRTResultYear)

LRTResultYearJSONImport = await FileAttachment("LRTResultYearJSON.json").json()

LRTResultByYearTable = Inputs.table(

LRTResultYearJSONImport,

// JSON.parse(LRTResultYear),

{columns: ["AE", "year", "ntrt", "nidot", "ndotj", "ndotdot", "logLR", "T_alpha", "pvalue", "alpha", "is_signal"],

width: {AE: 160},

// height: 550,

align: {

ntrt: "center",

nidot: "center",

ndotj: "center",

ndotdot: "center",

logLR: "center",

T_alpha: "center",

pvalue: "center",

is_signal: "center"

}

})

margin = ({top: 30, right: 30, bottom: 50, left: 240})

heatMapAE = addAnimation(

addTooltips(

Plot.plot({

padding: 0.01,

marginLeft: 250,

marks: [

Plot.cell(

LRTResultYearJSONImport,

// JSON.parse(LRTResultYear),

{

x: (d) => d.year,

y: (d) => d.AE,

fill: (d) => d.pvalue,

fillOpacity: 0.8,

inset: 0.5,

title: (d) =>

`${d.AE} \nYear: ${d.year} \nIs Signal: ${d.is_signal} \nLogLR: ${parseFloat(d.logLR).toFixed(2)} \nP-Value: ${d.pvalue} \nAlpha: ${d.alpha}`

}

),

Plot.text(

LRTResultYearJSONImport,

// JSON.parse(LRTResultYear),

{

x: "year",

y: "AE",

text: d => d.is_signal == true ? '❕' : ''

}

)

],

color: {

type: "linear",

domain: [0, 0.06, 1],

range: ["#fc836a", "#3BAFDA", "lightblue"]

},

x: {

axis: "top",

label: ''

},

y: {

label: ''

},

width,

height: 1500

})

),

{ type: "rect", delay: 2 }

)

Purpose-built for displays of data