pgenlibr Windows linking fix

2.0/pgenlibr/DESCRIPTION

@@ -2,7 +2,7 @@ Package: pgenlibr
 Type: Package
 Title: PLINK 2 Binary (.pgen) Reader
 Version: 0.3.6
-Date: 202x-xx-xx
+Date: 2024-04-30
 Authors@R: c(person(given = "Christopher",
                     family = "Chang",
                     role = c("aut", "cre"),
@@ -32,7 +32,7 @@ Authors@R: c(person(given = "Christopher",
                     comment = "Author of included Zstd library"))
 Description: A thin wrapper over PLINK 2's core libraries which provides an R
     interface for reading .pgen files.  A minimal .pvar loader is also
-    included.  Chang et al. (2015) <doi:10.1186/s13742-015-0047-8>.
+    included.  Chang et al. (2015) \doi{10.1186/s13742-015-0047-8}.
 License: LGPL (>= 3)
 Copyright: This package includes sources of the libdeflate library owned by
     Eric Biggers, sources of the SIMDe library owned by Evan Nemerson, sources

2.0/plink2.cc

@@ -72,7 +72,7 @@ static const char ver_str[] = "PLINK v2.00a6"
 #elif defined(USE_AOCL)
   " AMD"
 #endif
-  " (18 Apr 2024)";
+  " (28 Apr 2024)";
 static const char ver_str2[] =
   // include leading space if day < 10, so character length stays the same
   ""
@@ -864,7 +864,7 @@ PglErr Plink2Core(const Plink2Cmdline* pcp, MakePlink2Flags make_plink2_flags, c
       }
       // update (26 Nov 2017): change --no-pheno to also apply to .psam file
       const uint32_t ignore_psam_phenos = (!(pcp->fam_cols & kfFamCol6)) || (pcp->pheno_fname && pcp->pheno_range_list.name_ct);
-      reterr = LoadPsam(psamname, pcp->pheno_fname? nullptr : &(pcp->pheno_range_list), pcp->missing_catname, pcp->fam_cols, ignore_psam_phenos? 0 : 0x7fffffff, pcp->missing_pheno, (pcp->misc_flags / kfMiscAffection01) & 1, (pcp->misc_flags / kfMiscNoCategorical) & 1, pcp->max_thread_ct, &pii, &sample_include, &founder_info, &sex_nm, &sex_male, &pheno_cols, &pheno_names, &raw_sample_ct, &pheno_ct, &max_pheno_name_blen);
+      reterr = LoadPsam(psamname, pcp->pheno_fname? nullptr : &(pcp->pheno_range_list), pcp->missing_catname, pcp->fam_cols, ignore_psam_phenos? 0 : 0x7fffffff, pcp->missing_pheno, (pcp->misc_flags / kfMiscAffection01) & 1, (pcp->misc_flags / kfMiscNoCategorical) & 1, (pcp->misc_flags / kfMiscNeg9PhenoReallyMissing) & 1, pcp->max_thread_ct, &pii, &sample_include, &founder_info, &sex_nm, &sex_male, &pheno_cols, &pheno_names, &raw_sample_ct, &pheno_ct, &max_pheno_name_blen);
       if (unlikely(reterr)) {
         goto Plink2Core_ret_1;
       }
@@ -1161,7 +1161,7 @@ PglErr Plink2Core(const Plink2Cmdline* pcp, MakePlink2Flags make_plink2_flags, c
       }
     }
     if (pcp->pheno_fname) {
-      reterr = LoadPhenos(pcp->pheno_fname, &(pcp->pheno_range_list), sample_include, &pii.sii, pcp->missing_catname, raw_sample_ct, sample_ct, pcp->missing_pheno, (pcp->misc_flags / kfMiscAffection01) & 1, (pcp->misc_flags / kfMiscNoCategorical) & 1, (pcp->misc_flags / kfMiscPhenoIidOnly) & 1, (pcp->misc_flags / kfMiscPhenoColNums) & 1, pcp->max_thread_ct, &pheno_cols, &pheno_names, &pheno_ct, &max_pheno_name_blen);
+      reterr = LoadPhenos(pcp->pheno_fname, &(pcp->pheno_range_list), sample_include, &pii.sii, pcp->missing_catname, raw_sample_ct, sample_ct, pcp->missing_pheno, (pcp->misc_flags / kfMiscAffection01) & 1, (pcp->misc_flags / kfMiscNoCategorical) & 1, (pcp->misc_flags / kfMiscPhenoIidOnly) & 1, (pcp->misc_flags / kfMiscPhenoColNums) & 1, (pcp->misc_flags / kfMiscNeg9PhenoReallyMissing) & 1, pcp->max_thread_ct, &pheno_cols, &pheno_names, &pheno_ct, &max_pheno_name_blen);
       if (unlikely(reterr)) {
         goto Plink2Core_ret_1;
       }
@@ -1581,7 +1581,7 @@ PglErr Plink2Core(const Plink2Cmdline* pcp, MakePlink2Flags make_plink2_flags, c
       }
       if (pcp->covar_fname || pcp->covar_range_list.name_ct) {
         const char* cur_covar_fname = pcp->covar_fname? pcp->covar_fname : (pcp->pheno_fname? pcp->pheno_fname : psamname);
-        reterr = LoadPhenos(cur_covar_fname, &(pcp->covar_range_list), sample_include, &pii.sii, pcp->missing_catname, raw_sample_ct, sample_ct, pcp->missing_pheno, 2, (pcp->misc_flags / kfMiscNoCategorical) & 1, (pcp->misc_flags / kfMiscCovarIidOnly) & 1, (pcp->misc_flags / kfMiscCovarColNums) & 1, pcp->max_thread_ct, &covar_cols, &covar_names, &covar_ct, &max_covar_name_blen);
+        reterr = LoadPhenos(cur_covar_fname, &(pcp->covar_range_list), sample_include, &pii.sii, pcp->missing_catname, raw_sample_ct, sample_ct, pcp->missing_pheno, 2, (pcp->misc_flags / kfMiscNoCategorical) & 1, (pcp->misc_flags / kfMiscCovarIidOnly) & 1, (pcp->misc_flags / kfMiscCovarColNums) & 1, (pcp->misc_flags / kfMiscNeg9PhenoReallyMissing) & 1, pcp->max_thread_ct, &covar_cols, &covar_names, &covar_ct, &max_covar_name_blen);
         if (unlikely(reterr)) {
           goto Plink2Core_ret_1;
         }
@@ -6699,6 +6699,11 @@ int main(int argc, char** argv) {
             snprintf(g_logbuf, kLogbufSize, "Error: Invalid --input-missing-phenotype argument '%s' (must be an integer in [-2147483647, -1] or [3, 2147483647]).\n", cur_modif);
             goto main_ret_INVALID_CMDLINE_WWA;
           }
+          if (pc.missing_pheno == -9) {
+            // No reason to require --neg9-pheno-really-missing when user
+            // explicitly sets --input-missing-phenotype to -9.
+            pc.misc_flags |= kfMiscNeg9PhenoReallyMissing;
+          }
         } else if (strequal_k_unsafe(flagname_p2, "mport-dosage-certainty")) {
           if (unlikely(EnforceParamCtRange(argvk[arg_idx], param_ct, 1, 1))) {
             goto main_ret_INVALID_CMDLINE_2A;
@@ -8905,10 +8910,14 @@ int main(int argc, char** argv) {
           pc.fa_flags |= kfFaNormalize;
           pc.dependency_flags |= kfFilterPvarReq;
         } else if (strequal_k_unsafe(flagname_p2, "o-input-missing-phenotype")) {
-          if (pc.missing_pheno != -9) {
+          if (unlikely(pc.missing_pheno != -9)) {
             logerrputs("Error: --input-missing-phenotype and --no-input-missing-phenotype don't make\nsense together.\n");
             goto main_ret_INVALID_CMDLINE;
           }
+          if (unlikely(pc.misc_flags & kfMiscNeg9PhenoReallyMissing)) {
+            logerrputs("Error: --neg9-pheno-really-missing and --no-input-missing-phenotype don't make\nsense together.\n");
+            goto main_ret_INVALID_CMDLINE;
+          }
           pc.missing_pheno = 0;
           goto main_param_zero;
         } else if (strequal_k_unsafe(flagname_p2, "ative")) {
@@ -8935,6 +8944,9 @@ int main(int argc, char** argv) {
         } else if (strequal_k_unsafe(flagname_p2, "o-categorical")) {
           pc.misc_flags |= kfMiscNoCategorical;
           goto main_param_zero;
+        } else if (strequal_k_unsafe(flagname_p2, "eg9-pheno-really-missing")) {
+          pc.misc_flags |= kfMiscNeg9PhenoReallyMissing;
+          goto main_param_zero;
         } else if (likely(strequal_k_unsafe(flagname_p2, "o-id-header"))) {
           if (unlikely(EnforceParamCtRange(argvk[arg_idx], param_ct, 0, 1))) {
             goto main_ret_INVALID_CMDLINE_2A;

2.0/plink2_common.h

@@ -158,7 +158,8 @@ FLAGSET64_DEF_START()
   kfMiscMakeFoundersFirst = (1LLU << 46),
   kfMiscMakeFoundersNotfirst = (1LLU << 47),
   kfMiscMakeFoundersRequire2Missing = (1LLU << 48),
-  kfMiscYNosexMissingStats = (1LLU << 49)
+  kfMiscYNosexMissingStats = (1LLU << 49),
+  kfMiscNeg9PhenoReallyMissing = (1LLU << 50)
 FLAGSET64_DEF_END(MiscFlags);
 
 FLAGSET64_DEF_START()

2.0/plink2_help.cc

@@ -1872,12 +1872,21 @@ PglErr DispHelp(const char* const* argvk, uint32_t param_ct) {
               );
     // bugfix (27 Feb 2019): "\01" is interpreted as a single character, not a
     // null followed by a '1'...
-    HelpPrint("input-missing-phenotype\0no-input-missing-phenotype\0\061\0missing-catname\0missing-phenotype\0", &help_ctrl, 0,
+    HelpPrint("input-missing-phenotype\0no-input-missing-phenotype\0neg9-pheno-really-missing\0\061\0missing-catname\0missing-phenotype\0", &help_ctrl, 0,
 "  --input-missing-phenotype <v> : Set nonzero number to treat as a missing\n"
 "                                  pheno/covar in input files (default -9).\n"
 "  --no-input-missing-phenotype  : Don't treat any nonzero number as a missing\n"
 "                                  pheno/covar.  ('NA'/'nan' are still treated\n"
 "                                  as missing.)\n"
+"  --neg9-pheno-really-missing   : When neither --input-missing-phenotype nor\n"
+"                                  --no-input-missing-phenotype are specified,\n"
+"                                  and a pheno/covar value in [-8, -9) or\n"
+"                                  (-9, -10] is present, PLINK 2 now errors out\n"
+"                                  when -9 is also present: in this context it\n"
+"                                  is too likely that -9 does not represent a\n"
+"                                  missing value.\n"
+"                                  --neg9-pheno-really-missing suppresses the\n"
+"                                  error.\n"
 "  --1                           : Expect case/control phenotypes in input files\n"
 "                                  to be coded as 0 = control, 1 = case, instead\n"
 "                                  of the usual 0 = missing, 1 = ctrl, 2 = case.\n"

2.0/plink2_import.cc

@@ -3141,9 +3141,8 @@ PglErr VcfToPgen(const char* vcfname, const char* preexisting_psamname, const ch
         goto VcfToPgen_ret_MISSING_TOKENS;
       }
       if (unlikely(S_CAST(uintptr_t, id_end - pos_end) > kMaxIdBlen)) {
-        putc_unlocked('\n', stdout);
         snprintf(g_logbuf, kLogbufSize, "Error: Invalid ID on line %" PRIuPTR " of --vcf file (max " MAX_ID_SLEN_STR " chars).\n", line_idx);
-        goto VcfToPgen_ret_MALFORMED_INPUT_WW;
+        goto VcfToPgen_ret_MALFORMED_INPUT_WWN;
       }
 
       // note REF length
@@ -3153,6 +3152,10 @@ PglErr VcfToPgen(const char* vcfname, const char* preexisting_psamname, const ch
         goto VcfToPgen_ret_MISSING_TOKENS;
       }
       uint32_t cur_max_allele_slen = linebuf_iter - ref_allele_start;
+      if (unlikely(memchr(ref_allele_start, ',', cur_max_allele_slen) != nullptr)) {
+        snprintf(g_logbuf, kLogbufSize, "Error: Invalid REF allele on line %" PRIuPTR " of --vcf file.\n", line_idx);
+        goto VcfToPgen_ret_MALFORMED_INPUT_WWN;
+      }
 
       uint32_t alt_ct = 1;
       unsigned char ucc;
@@ -3870,6 +3873,10 @@ PglErr VcfToPgen(const char* vcfname, const char* preexisting_psamname, const ch
             goto VcfToPgen_load_start;
           }
           if ((!vic.vibc.gt_exists) && (!format_dosage_relevant) && (!format_hds_search)) {
+            line_iter = AdvToDelim(format_start, '\n');
+            if (unlikely(CountByte(format_start, '\t', line_iter - format_start) != sample_ct)) {
+              goto VcfToPgen_ret_MISSING_TOKENS;
+            }
             gparse_flags = kfGparseNull;
             genotext_byte_ct = 1;
           } else {
@@ -14868,7 +14875,7 @@ PglErr Plink1DosageToPgen(const char* dosagename, const char* famname, const cha
     uintptr_t* sex_male = nullptr;
     uintptr_t* founder_info = nullptr;
     uintptr_t max_pheno_name_blen = 0;
-    reterr = LoadPsam(famname, nullptr, missing_catname, fam_cols, 0x7fffffff, missing_pheno, (misc_flags / kfMiscAffection01) & 1, (misc_flags / kfMiscNoCategorical) & 1, max_thread_ct, &pii, &sample_include, &founder_info, &sex_nm, &sex_male, &pheno_cols, &pheno_names, &raw_sample_ct, &pheno_ct, &max_pheno_name_blen);
+    reterr = LoadPsam(famname, nullptr, missing_catname, fam_cols, 0x7fffffff, missing_pheno, (misc_flags / kfMiscAffection01) & 1, (misc_flags / kfMiscNoCategorical) & 1, (misc_flags / kfMiscNeg9PhenoReallyMissing) & 1, max_thread_ct, &pii, &sample_include, &founder_info, &sex_nm, &sex_male, &pheno_cols, &pheno_names, &raw_sample_ct, &pheno_ct, &max_pheno_name_blen);
     if (unlikely(reterr)) {
       goto Plink1DosageToPgen_ret_1;
     }

2.0/plink2_import_legacy.cc

@@ -635,7 +635,7 @@ PglErr RewritePsam(const char* in_psamname, const char* missing_catname, MiscFla
     uintptr_t* sex_male = nullptr;
     uintptr_t* founder_info = nullptr;
     uintptr_t max_pheno_name_blen = 0;
-    reterr = LoadPsam(in_psamname, nullptr, missing_catname, fam_cols, 0x7fffffff, missing_pheno, (misc_flags / kfMiscAffection01) & 1, (misc_flags / kfMiscNoCategorical) & 1, max_thread_ct, &pii, &sample_include, &founder_info, &sex_nm, &sex_male, &pheno_cols, &pheno_names, &raw_sample_ct, &pheno_ct, &max_pheno_name_blen);
+    reterr = LoadPsam(in_psamname, nullptr, missing_catname, fam_cols, 0x7fffffff, missing_pheno, (misc_flags / kfMiscAffection01) & 1, (misc_flags / kfMiscNoCategorical) & 1, (misc_flags / kfMiscNeg9PhenoReallyMissing) & 1, max_thread_ct, &pii, &sample_include, &founder_info, &sex_nm, &sex_male, &pheno_cols, &pheno_names, &raw_sample_ct, &pheno_ct, &max_pheno_name_blen);
     if (unlikely(reterr)) {
       goto RewritePsam_ret_1;
     }

2.0/plink2_psam.cc

@@ -42,7 +42,7 @@ typedef struct CatnameLl2Struct {
   char str[];
 } CatnameLl2;
 
-PglErr LoadPsam(const char* psamname, const RangeList* pheno_range_list_ptr, const char* missing_catname, FamCol fam_cols, uint32_t pheno_ct_max, int32_t missing_pheno, uint32_t affection_01, uint32_t no_categorical, uint32_t max_thread_ct, PedigreeIdInfo* piip, uintptr_t** sample_include_ptr, uintptr_t** founder_info_ptr, uintptr_t** sex_nm_ptr, uintptr_t** sex_male_ptr, PhenoCol** pheno_cols_ptr, char** pheno_names_ptr, uint32_t* raw_sample_ct_ptr, uint32_t* pheno_ct_ptr, uintptr_t* max_pheno_name_blen_ptr) {
+PglErr LoadPsam(const char* psamname, const RangeList* pheno_range_list_ptr, const char* missing_catname, FamCol fam_cols, uint32_t pheno_ct_max, int32_t missing_pheno, uint32_t affection_01, uint32_t no_categorical, uint32_t neg9_pheno_really_missing, uint32_t max_thread_ct, PedigreeIdInfo* piip, uintptr_t** sample_include_ptr, uintptr_t** founder_info_ptr, uintptr_t** sex_nm_ptr, uintptr_t** sex_male_ptr, PhenoCol** pheno_cols_ptr, char** pheno_names_ptr, uint32_t* raw_sample_ct_ptr, uint32_t* pheno_ct_ptr, uintptr_t* max_pheno_name_blen_ptr) {
   // outparameter pointers assumed to be initialized to nullptr
   //
   // pheno_ct_max should default to something like 0x7fffffff, not UINT32_MAX
@@ -726,7 +726,10 @@ PglErr LoadPsam(const char* psamname, const RangeList* pheno_range_list_ptr, con
     uintptr_t* founder_info = *founder_info_ptr;
     uintptr_t* sex_nm = *sex_nm_ptr;
     uintptr_t* sex_male = *sex_male_ptr;
+    const uint32_t check_neg8_to_10_pheno = (missing_pheno == -9) && (!neg9_pheno_really_missing);
     uint32_t sample_uidx = raw_sample_ct;
+    uint32_t neg9_seen = 0;
+    uint32_t other_neg8_to_10_seen = 0;
     while (sample_uidx) {
       --sample_uidx;
       unsigned char* cur_vardata = psam_info_reverse_ll->vardata;
@@ -746,6 +749,15 @@ PglErr LoadPsam(const char* psamname, const RangeList* pheno_range_list_ptr, con
               SetBit(sample_uidx, pheno_cols[pheno_idx].nonmiss);
               pheno_cols[pheno_idx].data.qt[sample_uidx] = dxx;
             }
+            if (check_neg8_to_10_pheno) {
+              if ((dxx <= -8.0) && (dxx >= -10.0)) {
+                if (dxx == -9.0) {
+                  neg9_seen = 1;
+                } else {
+                  other_neg8_to_10_seen = 1;
+                }
+              }
+            }
           } else {
             if (dxx == pheno_cased) {
               SetBit(sample_uidx, pheno_cols[pheno_idx].data.cc);
@@ -783,6 +795,13 @@ PglErr LoadPsam(const char* psamname, const RangeList* pheno_range_list_ptr, con
       }
       psam_info_reverse_ll = psam_info_reverse_ll->next;
     }
+    if (check_neg8_to_10_pheno && other_neg8_to_10_seen) {
+      if (unlikely(neg9_seen)) {
+        logerrputs("Error: Distinct phenotype values in [-8, -10] present, including -9, when -9 is\ntreated as missing.\n* Usually, this means -9 should be treated as an ordinary numeric value; use\n--no-input-missing-phenotype to specify this.\n* If -9 really does still mean missing, use --neg9-pheno-really-missing.\n");
+        goto LoadPsam_ret_INCONSISTENT_INPUT;
+      }
+      logerrputs("Warning: Phenotype value in [-8, -9) or (-9, -10] present, when -9 is treated\nas missing.  Use --no-input-missing-phenotype to treat -9 as a numeric value\n(missing values can be indicated by 'NA'), or --neg9-pheno-really-missing to\nsuppress this warning.\n");
+    }
     if (fid_present) {
       piip->sii.flags |= kfSampleIdFidPresent;
     }
@@ -856,7 +875,7 @@ typedef struct PhenoInfoLlStruct {
 
 // also for loading covariates.  set affection_01 to 2 to prohibit case/control
 // and make unnamed variables start with "COVAR" instead of "PHENO"
-PglErr LoadPhenos(const char* pheno_fname, const RangeList* pheno_range_list_ptr, const uintptr_t* sample_include, const SampleIdInfo* siip, const char* missing_catname, uint32_t raw_sample_ct, uint32_t sample_ct, int32_t missing_pheno, uint32_t affection_01, uint32_t no_categorical, uint32_t iid_only, uint32_t numeric_ranges, uint32_t max_thread_ct, PhenoCol** pheno_cols_ptr, char** pheno_names_ptr, uint32_t* pheno_ct_ptr, uintptr_t* max_pheno_name_blen_ptr) {
+PglErr LoadPhenos(const char* pheno_fname, const RangeList* pheno_range_list_ptr, const uintptr_t* sample_include, const SampleIdInfo* siip, const char* missing_catname, uint32_t raw_sample_ct, uint32_t sample_ct, int32_t missing_pheno, uint32_t affection_01, uint32_t no_categorical, uint32_t iid_only, uint32_t numeric_ranges, uint32_t neg9_pheno_really_missing, uint32_t max_thread_ct, PhenoCol** pheno_cols_ptr, char** pheno_names_ptr, uint32_t* pheno_ct_ptr, uintptr_t* max_pheno_name_blen_ptr) {
   unsigned char* bigstack_mark = g_bigstack_base;
   unsigned char* bigstack_end_mark = g_bigstack_end;
   char* pheno_names = nullptr;
@@ -1378,8 +1397,11 @@ PglErr LoadPhenos(const char* pheno_fname, const RangeList* pheno_range_list_ptr
           goto LoadPhenos_ret_NOMEM;
         }
       }
+      const uint32_t check_neg8_to_10_pheno = (missing_pheno == -9) && (!neg9_pheno_really_missing);
       const uintptr_t nonmiss_vec_ct = BitCtToVecCt(raw_sample_ct);
       uint32_t cat_pheno_idx = 0;
+      uint32_t neg9_seen = 0;
+      uint32_t other_neg8_to_10_seen = 0;
       PhenoCol* pheno_cols_iter = &(new_pheno_cols[old_pheno_ct]);
       for (uint32_t new_pheno_idx = 0; new_pheno_idx != new_pheno_ct; ++new_pheno_idx) {
         const uint32_t is_categorical = IsSet(categorical_phenos, new_pheno_idx);
@@ -1472,6 +1494,15 @@ PglErr LoadPhenos(const char* pheno_fname, const RangeList* pheno_range_list_ptr
                 SetBit(sample_uidx, pheno_cols_iter->nonmiss);
                 pheno_cols_iter->data.qt[sample_uidx] = dxx;
               }
+              if (check_neg8_to_10_pheno) {
+                if ((dxx <= -8.0) && (dxx >= -10.0)) {
+                  if (dxx == -9.0) {
+                    neg9_seen = 1;
+                  } else {
+                    other_neg8_to_10_seen = 1;
+                  }
+                }
+              }
             } else {
               if (dxx == pheno_cased) {
                 SetBit(sample_uidx, pheno_cols_iter->data.cc);
@@ -1485,6 +1516,13 @@ PglErr LoadPhenos(const char* pheno_fname, const RangeList* pheno_range_list_ptr
         }
         pheno_info_reverse_ll = pheno_info_reverse_ll->next;
       }
+      if (check_neg8_to_10_pheno && other_neg8_to_10_seen) {
+        if (unlikely(neg9_seen)) {
+          logerrputs("Error: Distinct phenotype/covariate values in [-8, -10] present, including -9,\nwhen -9 is treated as missing.  Use --no-input-missing-phenotype to treat -9 as\na numeric value (missing values can be indicated by 'NA'), or\n--neg9-pheno-really-missing to suppress this error.\n");
+          goto LoadPhenos_ret_INCONSISTENT_INPUT;
+        }
+        logerrputs("Warning: Phenotype/covariate value in [-8, -9) or (-9, -10] present, when -9\nis treated as missing.  Use --no-input-missing-phenotype to treat -9 as a\nnumeric value (missing values can be indicated by 'NA'), or\n--neg9-pheno-really-missing to suppress this warning.\n");
+      }
     }
     *pheno_names_ptr = pheno_names;
     *max_pheno_name_blen_ptr = max_pheno_name_blen;