Add dose-normalized exposure parameters

1library(yamaa)23yamaa_spec("spec.yaml")$run()$output

1import yamaa23yamaa.spec("spec.yaml").run().output

1# Reference solution for the yamaa benchmark adam-adpp-dose-normalized (R track).2#3# Written from the benchmark's full prompt and its inputs alone, as an agent4# would write it. Harbor's oracle agent runs it to check that the prompt can5# be solved and that the grader scores a correct answer 1.67library(dplyr, warn.conflicts = FALSE)8library(readr)910pp <- read_csv(11  "/app/input/pp.csv",12  col_types = cols(PPSTRESN = col_double(), .default = col_character())13)14ex <- read_csv(15  "/app/input/ex.csv",16  col_types = cols(EXDOSE = col_double(), .default = col_character())17)1819# The three exposure parameters that get a dose-normalized record.20normalized <- tibble(21  PPTESTCD = c("CMAX", "AUCLST", "AUCIFO"),22  NEWCD = c("CMAXD", "AUCLSTD", "AUCIFOD"),23  NEWPARAM = c(24    "Max Conc Norm by Dose",25    "AUC to Last Nonzero Conc Norm by Dose",26    "AUC Infinity Obs Norm by Dose"27  )28)29mg_per_unit <- tibble(EXDOSU = c("mg", "ug"), MGDIV = c(1, 1000))3031# Every PP record as collected.32kept <- pp |>33  transmute(34    STUDYID, USUBJID, VISIT,35    PARAMCD = PPTESTCD,36    PARAM = PPTEST,37    AVAL = PPSTRESN,38    AVALU = PPSTRESU39  )4041# The one administration that starts at each profile's reference time, its42# dose in mg (a dose in ug is a thousandth of a mg).43dose <- ex |>44  left_join(mg_per_unit, by = "EXDOSU") |>45  transmute(STUDYID, USUBJID, PPRFTDTC = EXSTDTC, DOSEMG = EXDOSE / MGDIV)4647added <- pp |>48  inner_join(normalized, by = "PPTESTCD") |>49  filter(!is.na(PPSTRESN)) |>50  inner_join(dose, by = c("STUDYID", "USUBJID", "PPRFTDTC")) |>51  filter(!is.na(DOSEMG), DOSEMG != 0) |>52  transmute(53    STUDYID, USUBJID, VISIT,54    PARAMCD = NEWCD,55    PARAM = NEWPARAM,56    AVAL = PPSTRESN / DOSEMG,57    AVALU = paste0(PPSTRESU, "/mg")58  )5960adpp <- bind_rows(kept, added) |>61  select(STUDYID, USUBJID, VISIT, PARAMCD, PARAM, AVAL, AVALU)6263dir.create("/app/output", showWarnings = FALSE)64write_csv(adpp, "/app/output/adpp.csv", na = "")

1# Reference solution for the yamaa benchmark adam-adpp-dose-normalized (Python track).2#3# Written from the benchmark's full prompt and its inputs alone, as an agent4# would write it. Harbor's oracle agent runs it to check that the prompt can5# be solved and that the grader scores a correct answer 1.67"""Create ADPP from EX and PP: collected PK parameters plus dose-normalized8records for CMAX, AUCLST, and AUCIFO, normalized by the dose (mg) of the9subject's administration starting at the profile's reference time."""1011import csv12import os1314INPUT = "/app/input"15OUTPUT = "/app/output"1617NORMALIZED = {18    "CMAX": ("CMAXD", "Max Conc Norm by Dose", "ng/mL/mg"),19    "AUCLST": ("AUCLSTD", "AUC to Last Nonzero Conc Norm by Dose", "h*ng/mL/mg"),20    "AUCIFO": ("AUCIFOD", "AUC Infinity Obs Norm by Dose", "h*ng/mL/mg"),21}2223COLUMNS = ["STUDYID", "USUBJID", "VISIT", "PARAMCD", "PARAM", "AVAL", "AVALU"]242526def read_csv(name):27    with open(os.path.join(INPUT, name), newline="") as f:28        return list(csv.DictReader(f))293031def dose_mg(ex_row):32    dose = float(ex_row["EXDOSE"])33    unit = ex_row["EXDOSU"].strip()34    if unit == "mg":35        return dose36    if unit in ("ug", "\u00b5g", "mcg"):37        return dose / 1000.038    return dose394041def main():42    ex = read_csv("ex.csv")43    pp = read_csv("pp.csv")4445    # One administration per subject per start time.46    dose_by_key = {}47    for row in ex:48        dose_by_key[(row["USUBJID"], row["EXSTDTC"])] = dose_mg(row)4950    out = []51    for rec in pp:52        base = {53            "STUDYID": rec["STUDYID"],54            "USUBJID": rec["USUBJID"],55            "VISIT": rec["VISIT"],56            "PARAMCD": rec["PPTESTCD"],57            "PARAM": rec["PPTEST"],58            "AVAL": rec["PPSTRESN"],59            "AVALU": rec["PPSTRESU"],60        }61        out.append(base)6263        spec = NORMALIZED.get(rec["PPTESTCD"])64        if spec is None or rec["PPSTRESN"] == "":65            continue66        dose = dose_by_key.get((rec["USUBJID"], rec["PPRFTDTC"]))67        if dose is None or dose == 0:68            continue69        paramcd, param, avalu = spec70        aval = float(rec["PPSTRESN"]) / dose71        added = dict(base)72        added["PARAMCD"] = paramcd73        added["PARAM"] = param74        added["AVAL"] = repr(aval)75        added["AVALU"] = avalu76        out.append(added)7778    os.makedirs(OUTPUT, exist_ok=True)79    with open(os.path.join(OUTPUT, "adpp.csv"), "w", newline="") as f:80        writer = csv.DictWriter(f, fieldnames=COLUMNS)81        writer.writeheader()82        writer.writerows(out)838485if __name__ == "__main__":86    main()

1/* opensas reference solution for the yamaa benchmark adam-adpp-dose-normalized.2   Reads the benchmark inputs and derives the requested output.3   Execute with the existing opensas runtime: opensas /app/output/result.sas. */45data pp;6  length STUDYID DOMAIN USUBJID PPSEQ PPTESTCD PPTEST PPCAT PPORRES PPORRESU PPSTRESC PPSTRESU PPSTAT PPREASND PPSPEC VISITNUM VISIT PPRFTDTC $1024;7  infile "/app/input/pp.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;8  input STUDYID : $char1024. DOMAIN : $char1024. USUBJID : $char1024. PPSEQ : $char1024. PPTESTCD : $char1024. PPTEST : $char1024. PPCAT : $char1024. PPORRES : $char1024. PPORRESU : $char1024. PPSTRESC : $char1024. PPSTRESN PPSTRESU : $char1024. PPSTAT : $char1024. PPREASND : $char1024. PPSPEC : $char1024. VISITNUM : $char1024. VISIT : $char1024. PPRFTDTC : $char1024.;9run;1011data ex;12  length STUDYID DOMAIN USUBJID EXSEQ EXTRT EXDOSU EXDOSFRM EXDOSFRQ EXROUTE EXSTDTC EXENDTC $1024;13  infile "/app/input/ex.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;14  input STUDYID : $char1024. DOMAIN : $char1024. USUBJID : $char1024. EXSEQ : $char1024. EXTRT : $char1024. EXDOSE EXDOSU : $char1024. EXDOSFRM : $char1024. EXDOSFRQ : $char1024. EXROUTE : $char1024. EXSTDTC : $char1024. EXENDTC : $char1024.;15run;1617/* Every PP record as collected. */18data kept;19  length STUDYID USUBJID VISIT PARAMCD PARAM AVALU $1024;20  set pp;21  PARAMCD=PPTESTCD; PARAM=PPTEST; AVAL=PPSTRESN; AVALU=PPSTRESU;22  keep STUDYID USUBJID VISIT PARAMCD PARAM AVAL AVALU;23run;2425/* The dose in mg of each administration; a dose in ug is a thousandth of a mg. */26data dose;27  length PPRFTDTC $1024;28  set ex;29  if EXDOSU='mg' then DOSEMG=EXDOSE;30  else if EXDOSU='ug' then DOSEMG=EXDOSE/1000;31  PPRFTDTC=EXSTDTC;32  keep STUDYID USUBJID PPRFTDTC DOSEMG;33run;3435/* The maximum concentration and area under the curve records that have a value. */36data exposure;37  length NEWCD NEWPARAM $1024;38  set pp;39  if PPTESTCD='CMAX' then do; NEWCD='CMAXD'; NEWPARAM='Max Conc Norm by Dose'; end;40  else if PPTESTCD='AUCLST' then do; NEWCD='AUCLSTD'; NEWPARAM='AUC to Last Nonzero Conc Norm by Dose'; end;41  else if PPTESTCD='AUCIFO' then do; NEWCD='AUCIFOD'; NEWPARAM='AUC Infinity Obs Norm by Dose'; end;42  if NEWCD ne '' and not missing(PPSTRESN);43run;4445/* The administration that starts at the profile's reference time. */46proc sql;47  create table matched as select a.*, b.DOSEMG from exposure as a inner join dose as b on a.STUDYID=b.STUDYID and a.USUBJID=b.USUBJID and a.PPRFTDTC=b.PPRFTDTC;48quit;4950data added;51  length STUDYID USUBJID VISIT PARAMCD PARAM AVALU $1024;52  set matched;53  if not missing(DOSEMG) and DOSEMG ne 0;54  PARAMCD=NEWCD; PARAM=NEWPARAM; AVAL=PPSTRESN/DOSEMG; AVALU=cats(PPSTRESU, '/mg');55  keep STUDYID USUBJID VISIT PARAMCD PARAM AVAL AVALU;56run;5758data result; set kept added; run;5960proc sql;61  create table final as select STUDYID, USUBJID, VISIT, PARAMCD, PARAM, AVAL, AVALU from result;62quit;63proc export data=final outfile="/app/output/adpp.csv" dbms=csv replace; run;