Aligator：路径分析 fortran 程序崩溃 "Segmentation fault"
Aligator: Pathway analysis fortran program crash with "Segmentation fault"
我尝试 运行 Aligator (http://x004.psycm.uwcm.ac.uk/~peter/) but it fails on my 64-bit Linux machine. When running aligator as stated in the document-file http://x004.psycm.uwcm.ac.uk/~peter/README_ALIGATOR.doc) 我得到这个错误:
./aligator < aligator.run > out.txt
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
#0 0x7FDCD5079577
#1 0x7FDCD5079B7E
#2 0x7FDCD457DCAF
#3 0x401DBF in MAIN__ at aligator.f:?
Segmentation fault
我觉得这个 fortran 软件是为 32 位系统编写的，这就是为什么我也必须像这样编译 aligator.f:
f77 aligator.f -mcmodel=large -o aligator
aligator.f的源代码：
CHARACTER*50 INFILE,OUTFILE
CHARACTER*8 PROCNAM(5),PXN
INTEGER JSNP(6000000,25),NSNP(6000000)
INTEGER ISNP(6000000)
INTEGER ILIS(100,2)
INTEGER NGS(1000000),IDSNP(1000000)
INTEGER IGS(1000000,25)
INTEGER ICH(20000),IGOC(20000)
INTEGER IDGE(30000),IPFC(35000,2)
INTEGER IDGO(20000),NGC(30000)
INTEGER JDGO(20000),MGC(30000)
INTEGER MPFC(20000),MCG(30000),NCG(30000)
INTEGER IGC(30000,1000),JSCO(20000)
INTEGER JGC(30000,1000),ISGSNP(200000)
INTEGER KSCO(50000,10000),MSCO(20000)
INTEGER NUMSIG(3,5)
INTEGER MUMSIG(3,5),ISLIM(200)
INTEGER ISCO(20000),MUM(20000),NUN(20000)
REAL PR(20000),CRIT(3),CUM(1000001),QR(20000)
REAL ZR(20000),YR(3,5),GM(20000),QZ(5),PMIN(5)
REAL DUM(1000001),QMIN(5)
REAL*8 QP(20000),CHX,RU(20000),QQ(20000)
REAL*8 XX,ST,YY
REAL*8 GETRAN
EXTERNAL GETRAN,QUIKSORT
OPEN(10,FILE='allsnp_gene_NCBI_0KB.txt')
C THIS FILE CONTAINS A LIST OF ALL SNPs ANNOTATED
C BY THE GENES THEY LIE WITHIN (NCBI ID NUMBERS)
OPEN(25,FILE='allsnp_gene_NCBI_0KB.list')
OPEN(11,FILE='gene_go_17_03_09.dat')
OPEN(12,FILE='pfc_17_03_09.dat')
READ(*,*) INFILE
READ(*,*) OUTFILE
OPEN(14,FILE=INFILE)
OPEN(15,FILE=OUTFILE)
READ(*,*) NSNPLIST
READ(*,*) DRIT
READ(*,*) NWIN
READ(*,*) NREP
READ(*,*) ST
NSG=5770877
C TOTAL NUMBER OF SNPS IN FILE allsnp_gene_NCBI_0KB.txt
NGO=754498
C TOTAL NUMBER OF GENE/GO CATEGORY ASSIGNMENTS
NPFC=28339
C TOTAL NUMBER OF GO CATEGORIES WITH NAMES/FUNCTION DATA
PROCNAM(1)='ALL CATS'
PROCNAM(2)='CELLULAR'
PROCNAM(3)='FUNCTION'
PROCNAM(4)='PROCESS '
PROCNAM(5)='PRO+FUNC'
CRIT(1)=0.05
CRIT(2)=0.01
CRIT(3)=0.001
NUM=0
DO 873 I=1,NPFC
READ(12,*) (IPFC(I,K),K=1,2)
873 CONTINUE
DO 871 I=1,1000000
NGS(I)=0
871 CONTINUE
DO 872 I=1,30000
MGC(I)=0
MCG(I)=0
872 CONTINUE
DO 604 I=10,99
READ(25,*) II,ILIS(I,1),ILIS(I,2)
604 CONTINUE
C READING IN START AND END POINTS ON THE COMPLETE LIST OF SNPS WHOSE
C RS NUMBERS START WITH 10, 11, .... 99
C TO MAKE IT FASTER TO LOCATE THE SNPS USED IN A PARTICULAR STUDY.
MMSP=0
DO 503 I=1,NSG
READ(10,*) ISNP(I),NSNP(I),(JSNP(I,K),K=1,NSNP(I))
503 CONTINUE
C READS IN COMPLETE LIST OF SNPS WITHIN GENES
MGE=0
MSP=0
DO 605 I=1,NSNPLIST
C THIS SECTION READS IN THE SNPS USED IN THE STUDY BEING ANALYSED
C CHECKS THEM AGAINST THE TOTAL LIST OF SNPS WITHIN GENES
C TO CREATE A LIST OF SNPS WITHIN GENES AND THE GENES WITHIN WHICH THEY LIE
READ(14,*) II,PP
RS=REAL(II)
RX=LOG10(RS)
IX=INT(RX)-1
RJ=REAL(IX)
IC=INT(RS/(10**RJ))
IUS=0
DO 696 J=ILIS(IC,1),ILIS(IC,2)
IF (II.EQ.ISNP(J)) THEN
IUS=J
GOTO 697
ENDIF
696 CONTINUE
697 CONTINUE
IF (IUS.EQ.0) THEN
GOTO 605
ENDIF
MSP=MSP+1
IDSNP(MSP)=ISNP(IUS)
NGS(MSP)=NSNP(IUS)
DO 698 J=1,NSNP(IUS)
JGX=JSNP(IUS,J)
JNE=0
DO 802 II=1,MGE
IF (JGX.EQ.IDGE(II)) THEN
JNE=II
ENDIF
802 CONTINUE
IF (JNE.EQ.0) THEN
MGE=MGE+1
IDGE(MGE)=JGX
JNE=MGE
ENDIF
IGS(MSP,J)=JNE
698 CONTINUE
IF (PP.LT.DRIT) THEN
NUM=NUM+1
ISGSNP(NUM)=MSP
ENDIF
C CREATES LIST OF SIGNIFICANT SNPS WITH P-val < SPECIFIED CRITERION
699 CONTINUE
605 CONTINUE
DUM(1)=0.0
DO 603 I=1,MSP
DUM(I+1)=DUM(I)+(1.0/FLOAT(MSP))
603 CONTINUE
DUM(MSP+1)=1.0
PRINT*,'SNPS READ'
PRINT*,'#SNPS = ',MSP
PRINT*,'#GENES = ',MGE
DO 502 I=1,NGO
READ(11,*) JGX,IID
INE=0
DO 842 II=1,MGE
IF (JGX.EQ.IDGE(II)) THEN
INE=II
GOTO 937
ENDIF
842 CONTINUE
937 CONTINUE
IF (INE.EQ.0) THEN
GOTO 502
ENDIF
JNE=0
DO 602 II=1,MMGO
IF (IID.EQ.JDGO(II)) THEN
JNE=II
ENDIF
602 CONTINUE
IF (JNE.EQ.0) THEN
MMGO=MMGO+1
C PRINT*,MMGO,IID
JDGO(MMGO)=IID
JNE=MMGO
ENDIF
INF=0
DO 722 J=1,MGC(INE)
IF (JNE.EQ.JGC(INE,J)) THEN
INF=J
ENDIF
722 CONTINUE
IF (INF.EQ.0) THEN
MGC(INE)=MGC(INE)+1
MCG(JNE)=MCG(JNE)+1
JGC(INE,MGC(INE))=JNE
ENDIF
C JGC CONTAINS ALL GO GATEGORY MEMBERSHIPS FOR EACH GENE
502 CONTINUE
MGO=0
DO 607 I=1,MMGO
C THIS SECTION REMOVES GO CATEGORIES WITH FEWER THAN 3 GENES
C OR MORE THAN 5000 (USER CAN EDIT AND RECOMPILE IF DESIRED)
IF (MCG(I).LT.3.OR.MCG(I).GT.5000) THEN
ICH(I)=0
GOTO 607
ENDIF
MGO=MGO+1
NCG(MGO)=MCG(I)
IDGO(MGO)=JDGO(I)
ICH(I)=MGO
IPF=0
DO 874 J=1,NPFC
IF (IDGO(MGO).EQ.IPFC(J,1)) THEN
IPF=J
GOTO 931
ENDIF
874 CONTINUE
931 CONTINUE
IF (IPF.EQ.0) THEN
PRINT*,'MISSING CATEGORY ',IDGO(MGO)
ELSE
MPFC(MGO)=IPFC(IPF,2)
ENDIF
607 CONTINUE
DO 609 I=1,MGE
NGC(I)=0
NGC(I)=0
DO 608 J=1,MGC(I)
JX=JGC(I,J)
NN=ICH(JX)
IF (NN.NE.0) THEN
NGC(I)=NGC(I)+1
IGC(I,NGC(I))=NN
ENDIF
608 CONTINUE
609 CONTINUE
PRINT*,'CATS READ'
PRINT*,'No.CATS = ',MGO
NMA=0
JMA=0
DO 226 I=1,MGE
IF (NGC(I).GE.NMA) THEN
NMA=NGC(I)
JMA=I
ENDIF
226 CONTINUE
PRINT*,'MAX #CATS PER GENE = ',NMA
PRINT*,'GENE ID ',IDGE(JMA)
NMA=0
JMA=0
DO 2 I=1,MSP
IF (NGS(I).GE.NMA) THEN
NMA=NGS(I)
JMA=I
ENDIF
2 CONTINUE
PRINT*,'NO. SNPS IN STUDY = ',NSNPLIST
PRINT*,'NO. SNPS IN GENES = ',MSP
PRINT*,'MAX #GENES PER SNP = ',NMA
PRINT*,'SNP ID ',IDSNP(JMA)
NMA=0
JMA=0
DO 3 I=1,MGO
IF (NGC(I).GE.NMA) THEN
NMA=NGC(I)
JMA=I
ENDIF
3 CONTINUE
DO 987 IY=1,20000
ISCO(IY)=0
JSCO(IY)=0
987 CONTINUE
DO 587 IY=1,3
DO 787 IZ=1,5
NUMSIG(IY,IZ)=0
787 CONTINUE
587 CONTINUE
JGE=0
DO 14 I=1,NUM
DO 198 J=1,20000
IGOC(J)=0
198 CONTINUE
KU=ISGSNP(I)
C GET THE ID OF EACH SIGNIFICANT SNP
DO 16 J=1,NGS(KU)
ICG=IGS(KU,J)
C WHAT GENES DOES IT LIE IN ?
IUS=0
DO 26 K=1,JGE
IF (ICG.EQ.JSCO(K)) THEN
IUS=K
ENDIF
26 CONTINUE
C IS THIS GENE ALREADY ON THE LIST ?
IF (IUS.EQ.0) THEN
JGE=JGE+1
JSCO(JGE)=ICG
C ADD GENE TO LIST
DO 18 KK=1,NGC(ICG)
K=IGC(ICG,KK)
IF (IGOC(K).EQ.0) THEN
ISCO(K)=ISCO(K)+1
IGOC(K)=1
ENDIF
C INCREASE THE COUNT OF SIGNIFICANT GENES FOR ALL GO CATEGORIES
C CONTAINING THIS GENE BY 1 - BUT ONLY IF THIS SNP HASN'T ALREADY
C CONTRIBUTED A GENE TO THIS CATEGORY. IF A SNP LIES
C IN SEVERAL OVERLAPPING GENES ALL IN THE SAME GO CATEGORY,
C THIS PREVENTS ALL THE GENES BEING COUNTED SEPARATELY.
18 CONTINUE
ENDIF
16 CONTINUE
14 CONTINUE
PRINT*,'P-VALUE CRITERION FOR DEFINING SIGNIFICANT SNPS = ',DRIT
PRINT*,'NUMBER OF GENES WITH SIGNIFICANT SNPS = ',JGE
PRINT*,'NUMBER OF SIMULATED REPLICATE GENE-LISTS = ',NWIN
PRINT*,'NUMBER OF BOOTSTRAP STUDIES = ',NREP
DO 979 IT=1,NWIN
PRINT*,'SIMULATED REPLICATE GENELIST',IT
KRE=1
DO 887 IY=1,10000
JSCO(IY)=0
GM(IY)=0
KSCO(IT,IY)=0
887 CONTINUE
KGE=0
666 CONTINUE
YY=GETRAN(XX,ST)
R=REAL(XX)
KU=0
DO 551 K=1,MSP
IF (R.GE.DUM(K).AND.R.LT.DUM(K+1)) THEN
KU=K
ENDIF
551 CONTINUE
C SELECT A RANDOM SNP
DO 298 J=1,20000
IGOC(J)=0
298 CONTINUE
DO 116 J=1,NGS(KU)
ICG=IGS(KU,J)
C WHAT GENES DOES THE SNP LIE IN ?
IUS=0
DO 216 K=1,KGE
IF (ICG.EQ.JSCO(K)) THEN
IUS=K
ENDIF
216 CONTINUE
C GENE ALREADY ON LIST ?
C PRINT*,II,NID-I,IDJ,IUS,ICG
IF (IUS.EQ.0) THEN
KGE=KGE+1
C PRINT*,IT,KGE
IF (KGE.GT.JGE) THEN
GOTO 994
ENDIF
C GENELIST REACHED THE SAME LENGTH AS THE ORIGINAL ?
JSCO(KGE)=ICG
DO 118 KK=1,NGC(ICG)
K=IGC(ICG,KK)
IF (IGOC(K).EQ.0) THEN
KSCO(IT,K)=KSCO(IT,K)+1
IGOC(K)=1
ENDIF
118 CONTINUE
C ADD 1 TO EACH GO CATEGORY CONTAINING THE GENE (SEE ABOVE)
ENDIF
116 CONTINUE
GOTO 666
994 CONTINUE
979 CONTINUE
DO 812 IX=1,5
PMIN(IX)=99.0
812 CONTINUE
NSCA=0
DO 22 I=1,MGO
NMP=MPFC(I)
IF (ISCO(I).LE.1) THEN
GOTO 22
ENDIF
C ONLY COUNT GO CATEGORIES WITH TWO OR MORE SIGNIFICANT GENES
PR(I)=0
ZR(I)=0
RU(I)=0.0d0
CUM(1)=0
DO 32 J=1,NWIN
IF (ISCO(I).LE.KSCO(J,I)) THEN
PR(I)=PR(I)+(1.0/FLOAT(NWIN))
ENDIF
C CATEGORY-SPECIFIC P-VALUE = NUMBER OF SIMULATED GENELISTS
C WHERE THE SCORE FOR THE CATEGORY (KSCO) IS AT LEAST AS HIGH
C AS THAT IN THE REAL DATA (ISCO)
CUM(J+1)=CUM(J)+(1.0/FLOAT(NWIN))
GM(I)=GM(I)+(FLOAT(KSCO(J,I))/FLOAT(NWIN))
C CALCULATING EXPECTED NUMBER OF SIGNIFICANT GENES IN THE CATEGORY
32 CONTINUE
CUM(NWIN+1)=1.0
IF (PR(I).LT.PMIN(1)) THEN
PMIN(1)=PR(I)
ENDIF
DO 813 J=1,3
IF (NMP.EQ.J.AND.PR(I).LE.PMIN(J+1)) THEN
PMIN(J+1)=PR(I)
ENDIF
813 CONTINUE
IF (NMP.GE.2.AND.PR(I).LE.PMIN(5)) THEN
PMIN(5)=PR(I)
ENDIF
C CALCULATING MINIMUM P-VALUES FOR CATEGORIES OF EACH FUNCTIONAL TYPE
DO 33 J=1,3
IF (PR(I).LE.CRIT(J)) THEN
NUMSIG(J,1)=NUMSIG(J,1)+1
NUMSIG(J,NMP+1)=NUMSIG(J,NMP+1)+1
IF (NMP.GE.2) THEN
NUMSIG(J,5)=NUMSIG(J,5)+1
ENDIF
ENDIF
C COUNTING NUMBER OF CATEGORIES REACHING A GIVEN CATEGORY-SPECIFIC P-VALUE
33 CONTINUE
QP(I)=DBLE(PR(I))
22 CONTINUE
NBOOT=NWIN
DO 716 IC=1,3
DO 816 ID=1,5
YR(IC,ID)=0
816 CONTINUE
716 CONTINUE
DO 819 I=1,5
QZ(I)=0
819 CONTINUE
DO 51 I=1,NREP
C PRINT*,I
C PERFORMING REPLICATE "STUDIES" TO GET SIGNIFICANCE OF NUMBERS OF
C SIGNIFICANTLY OVER-REPRESENTED CATEGORIES
YY=GETRAN(XX,ST)
R=REAL(XX)
IYS=0
DO 53 K=1,NWIN
IF (R.GE.CUM(K).AND.R.LT.CUM(K+1)) THEN
IYS=K
ENDIF
53 CONTINUE
C CHOOSE A REPLICATE GENE-LIST TO ACT AS THE "REAL DATA"
DO 54 K=1,MGO
MSCO(K)=KSCO(IYS,K)
QR(K)=0
54 CONTINUE
DO 58 K=1,3
DO 589 KK=1,5
MUMSIG(K,KK)=0
589 CONTINUE
58 CONTINUE
DO 52 J=1,NBOOT
528 CONTINUE
YY=GETRAN(XX,ST)
R=REAL(XX)
JYS=0
DO 55 K=1,NWIN
IF (R.GE.CUM(K).AND.R.LT.CUM(K+1)) THEN
JYS=K
ENDIF
55 CONTINUE
IF (JYS.EQ.IYS) THEN
GOTO 528
ENDIF
C SELECT REPLICATE GENE LISTS (WITH REPLACEMENT)
DO 56 K=1,MGO
IF (MSCO(K).LE.KSCO(JYS,K)) THEN
QR(K)=QR(K)+(1.0/FLOAT(NBOOT))
ENDIF
C OBTAIN CATEGORY-SPECIFIC P-VALUES FOR THE "REAL DATA"
56 CONTINUE
52 CONTINUE
DO 814 K=1,5
QMIN(K)=99.0
814 CONTINUE
DO 57 K=1,MGO
NMP=MPFC(K)
IF (MSCO(K).LE.1) THEN
GOTO 57
ENDIF
JGG=IDGO(K)
IF (QR(K).LT.QMIN(1)) THEN
QMIN(1)=QR(K)
ENDIF
DO 815 JX=1,3
IF (NMP.EQ.JX.AND.QR(K).LE.QMIN(JX+1)) THEN
QMIN(JX+1)=QR(K)
ENDIF
815 CONTINUE
IF (NMP.GE.2.AND.QR(K).LE.QMIN(5)) THEN
QMIN(5)=QR(K)
ENDIF
C OBTAIN MINIMUM P-VALUES FOR CATEGORIES OF VARIOUS FUNCTIONAL TYPES
DO 63 JJ=1,3
IF (QR(K).LE.CRIT(JJ)) THEN
MUMSIG(JJ,1)=MUMSIG(JJ,1)+1
MUMSIG(JJ,NMP+1)=MUMSIG(JJ,NMP+1)+1
IF (NMP.GE.2) THEN
MUMSIG(JJ,5)=MUMSIG(JJ,5)+1
ENDIF
ENDIF
C OBTAIN NUMBERS OF CATEGORIES WITH CATEGORY-SPECIFIC P-VALUES LESS THAN
C VARIOUS CRITERIA
63 CONTINUE
57 CONTINUE
DO 67 K=1,MGO
IF (ISCO(K).LE.1) THEN
GOTO 67
ENDIF
IF (PR(K).GE.QMIN(1)) THEN
ZR(K)=ZR(K)+(1.0/FLOAT(NREP))
ENDIF
C PROBABILITY THAT ANY CATEGORY-SPECIFIC P-VALUE IN A STUDY WHERE THERE
C IS NO FUNCTIONAL ASSOCIATION (i.e. RANDOMLY-SELECTED "REAL DATA") IS
C LESS THAN OR EQUAL TO THE P-VALUE OF A GIVEN CATEGORY IN THE ORIGINAL
C DATASET (i.e. A STUDY-WIDE P-VALUE CORRECTED FOR TESTING MULTIPLE NON-
C INDEPENDENT CATEGORIES)
DO 77 KK=1,MGO
IF (MSCO(KK).LE.1) THEN
GOTO 77
ENDIF
IF (PR(K).GE.QR(KK)) THEN
RU(K)=RU(K)+(1.0d0/DFLOAT(NREP))
ENDIF
C EXPECTED NUMBER OF CATEGORIES PER STUDY WITH CATEGORY-SPECIFIC P-vALUES
C LESS THAN OR EQUAL TO THAT OF A GIVEN CATEGORY IN THE ORIGINAL DATASET
77 CONTINUE
67 CONTINUE
DO 817 JJ=1,5
IF (QMIN(JJ).LE.PMIN(JJ)) THEN
QZ(JJ)=QZ(JJ)+(1.0/FLOAT(NREP))
ENDIF
C PROBABILITY THAT THE MINIMUM CATEGORY-SPECIFIC P-VALUE IN A STUDY WHERE THERE
C IS NO FUNCTIONAL ASSOCIATION (i.e. RANDOMLY-SELECTED "REAL DATA") IS
C LESS THAN OR EQUAL TO THE MINIMUM CATEGORY-SPECIFIC P-VALUE IN THE ORIGINAL
C DATA
817 CONTINUE
DO 64 JJ=1,3
DO 642 JH=1,5
IF (MUMSIG(JJ,JH).GE.NUMSIG(JJ,JH)) THEN
YR(JJ,JH)=YR(JJ,JH)+(1.0/FLOAT(NREP))
ENDIF
C PROBABILITY THAN THE NUMBER OF CATEGORIES REACHING SIGNIFICANCE IN A STUDY
C WHERE THERE IS NO FUNCTIONAL ASSOCIATION (i.e. RANDOMLY-SELECTED "REAL DATA")
C IS LESS THAN OR EQUAL TO THAT IN THE ORIGINAL DATA (i.e. TESTING FOR A
C SIGNIFICANTLY LARGE NUMBER OF OVER-REPRESENTED CATEGORIES)
642 CONTINUE
64 CONTINUE
51 CONTINUE
DO 712 IT=1,MGO
IF (ISCO(IT).EQ.0) THEN
QP(IT)=1d0+DBLE(GM(IT))
ENDIF
712 CONTINUE
X=QUIKSORT(QP,MGO,MUM)
DO 23 I=1,MGO
J=MUM(I)
NMP=MPFC(J)
PXN=PROCNAM(NMP+1)
JG=IDGO(J)
JH=ISCO(J)
IF (ISCO(J).LE.1) THEN
GOTO 23
ENDIF
C ONLY COUNT GO CATEGORIES WITH TWO OR MORE SIGNIFICANT GENES
WRITE(15,44) JG,PXN,JH,JGE,NCG(J),MGE,GM(J),PR(J),ZR(J),RU(J)
23 CONTINUE
WRITE(15,*)
WRITE(15,*)
DO 691 ISS=1,5
WRITE(15,*) PROCNAM(ISS)
WRITE(15,*)
WRITE(15,*) 'MIN P =',PMIN(ISS),' SIG = ',QZ(ISS)
WRITE(15,*)
DO 69 IS=1,3
WRITE(15,*) 'NO. OF P<',CRIT(IS),' =',NUMSIG(IS,ISS)
WRITE(15,*) 'SIG = ',YR(IS,ISS)
69 CONTINUE
WRITE(15,*)
691 CONTINUE
CLOSE(15)
44 FORMAT(I8,2X,A8,3(2X,I5),2X,I5,2X,F7.2,2X,F7.5,2X,F6.4,2X,F9.2)
STOP
END
REAL*8 FUNCTION GETRAN(X,S)
REAL*8 X,S,A,B,SN,SS,RN
INTEGER IS
A=7D0**5
B=(2D0**31)-1
SN=S*A
IS=INT(SN/B)
SS=DFLOAT(IS)
C PRINT*,SN,SS,IS
RN=SN-(SS*B)
S=RN
X=S/B
GETRAN=1D0
RETURN
END
REAL FUNCTION QUIKSORT(RLOD,NLO,NUM)
REAL*8 RLOD(20000)
REAL*8 SORT1(20000),SORT2(20000),SORT3(20000)
INTEGER NUM(20000),NUM1(20000),NUM2(20000)
DO 5 I=1,NLO
SORT1(I)=-1000.0D0
NUM1(I)=0
5 CONTINUE
DO 10 I=1,NLO
XLO=RLOD(I)
C PRINT*,XLO
DO 20 J=NLO+1-I,NLO
IF (XLO.GE.SORT1(J)) THEN
C PRINT*,XLO,SORT1(J),J
IPOS=J
ENDIF
20 CONTINUE
C PRINT*,IPOS
SORT2(IPOS)=XLO
NUM2(IPOS)=I
DO 30 K=2,IPOS
SORT2(K-1)=SORT1(K)
NUM2(K-1)=NUM1(K)
30 CONTINUE
DO 40 K=IPOS+1,NLO
SORT2(K)=SORT1(K)
NUM2(K)=NUM1(K)
40 CONTINUE
DO 50 L=1,NLO
NUM1(L)=NUM2(L)
SORT1(L)=SORT2(L)
C PRINT*,L,SORT1(L)
50 CONTINUE
10 CONTINUE
DO 60 I=1,NLO
NUM(I)=NUM1(I)
60 CONTINUE
QUIKSORT=1.0
RETURN
END
因为我不是fortran-coder，不知道这段代码能不能改成完全支持64位系统？
编辑：
我试过这样编译：
gfortran aligator.f -o aligator -g -fcheck=all -fbacktrace
有这些错误：
/tmp/ccbJsCPS.o: In function `MAIN__':
/home/alexsson/Programs/Aligator/aligator.f:354:(.text+0x628a): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:355:(.text+0x62ff): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:356:(.text+0x6374): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:357:(.text+0x6380): relocation truncated to fit: R_X86_64_PC32 against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:360:(.text+0x65c5): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:360:(.text+0x65eb): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:365:(.text+0x66c3): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:365:(.text+0x66e9): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:369:(.text+0x697d): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:370:(.text+0x69ee): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:371:(.text+0x6a6a): additional relocation overflows omitted from the output
collect2: error: ld returned 1 exit status
我不确定我是否变得更聪明了...
解法：
感谢 tkoenig，解决方案是将 aligator.f 的源代码更改为：http://pastebin.com/tnYhuba3
它适用于我的 Slackware 64 位当前版本（已启用 multilib）- 具有 16 GB RAM 的系统。
你有一些真的 大数组，这会导致 32 位寻址模式出现问题。如果您使数组可分配并在程序开始时分配它们，就像 with
CHARACTER*50 INFILE,OUTFILE
CHARACTER*8 PROCNAM(5),PXN
integer, dimension(:,:), allocatable :: jsnp
integer, dimension(:), allocatable :: nsnp, isnp
INTEGER ILIS(100,2)
INTEGER NGS(1000000),IDSNP(1000000)
INTEGER IGS(1000000,25)
INTEGER ICH(20000),IGOC(20000)
INTEGER IDGE(30000),IPFC(35000,2)
INTEGER IDGO(20000),NGC(30000)
INTEGER JDGO(20000),MGC(30000)
INTEGER MPFC(20000),MCG(30000),NCG(30000)
integer, dimension(:,:), allocatable :: igc, jgc, ksco
INTEGER JSCO(20000)
INTEGER ISGSNP(200000)
INTEGER MSCO(20000)
INTEGER NUMSIG(3,5)
INTEGER MUMSIG(3,5),ISLIM(200)
INTEGER ISCO(20000),MUM(20000),NUN(20000)
REAL PR(20000),CRIT(3),CUM(1000001),QR(20000)
REAL ZR(20000),YR(3,5),GM(20000),QZ(5),PMIN(5)
REAL DUM(1000001),QMIN(5)
REAL*8 QP(20000),CHX,RU(20000),QQ(20000)
REAL*8 XX,ST,YY
REAL*8 GETRAN
EXTERNAL GETRAN,QUIKSORT
allocate(JSNP(6000000,25))
allocate(NSNP(6000000), ISNP(6000000))
allocate(IGC(30000,1000),JGC(30000,1000),KSCO(50000,10000))
这应该不再是一个问题。
内存不足，无法查看。
根据您需要为 INTEGER 使用的值的范围，将默认整数更改为两字节或一字节的类型可能也很有趣，如
integer, parameter :: i2 = selected_int_kind(4) ! select range of at least 10000, usually translates into two bytes
integer(kind=i2), dimension(:,:), allocatable :: jsnp
或
integer, parameter :: i1 = selected_int_kind(2) ! one-byte integer
这可以显着节省内存和速度。
我尝试 运行 Aligator (http://x004.psycm.uwcm.ac.uk/~peter/) but it fails on my 64-bit Linux machine. When running aligator as stated in the document-file http://x004.psycm.uwcm.ac.uk/~peter/README_ALIGATOR.doc) 我得到这个错误:
./aligator < aligator.run > out.txt
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
#0 0x7FDCD5079577
#1 0x7FDCD5079B7E
#2 0x7FDCD457DCAF
#3 0x401DBF in MAIN__ at aligator.f:?
Segmentation fault
我觉得这个 fortran 软件是为 32 位系统编写的，这就是为什么我也必须像这样编译 aligator.f:
f77 aligator.f -mcmodel=large -o aligator
aligator.f的源代码：
CHARACTER*50 INFILE,OUTFILE
CHARACTER*8 PROCNAM(5),PXN
INTEGER JSNP(6000000,25),NSNP(6000000)
INTEGER ISNP(6000000)
INTEGER ILIS(100,2)
INTEGER NGS(1000000),IDSNP(1000000)
INTEGER IGS(1000000,25)
INTEGER ICH(20000),IGOC(20000)
INTEGER IDGE(30000),IPFC(35000,2)
INTEGER IDGO(20000),NGC(30000)
INTEGER JDGO(20000),MGC(30000)
INTEGER MPFC(20000),MCG(30000),NCG(30000)
INTEGER IGC(30000,1000),JSCO(20000)
INTEGER JGC(30000,1000),ISGSNP(200000)
INTEGER KSCO(50000,10000),MSCO(20000)
INTEGER NUMSIG(3,5)
INTEGER MUMSIG(3,5),ISLIM(200)
INTEGER ISCO(20000),MUM(20000),NUN(20000)
REAL PR(20000),CRIT(3),CUM(1000001),QR(20000)
REAL ZR(20000),YR(3,5),GM(20000),QZ(5),PMIN(5)
REAL DUM(1000001),QMIN(5)
REAL*8 QP(20000),CHX,RU(20000),QQ(20000)
REAL*8 XX,ST,YY
REAL*8 GETRAN
EXTERNAL GETRAN,QUIKSORT
OPEN(10,FILE='allsnp_gene_NCBI_0KB.txt')
C THIS FILE CONTAINS A LIST OF ALL SNPs ANNOTATED
C BY THE GENES THEY LIE WITHIN (NCBI ID NUMBERS)
OPEN(25,FILE='allsnp_gene_NCBI_0KB.list')
OPEN(11,FILE='gene_go_17_03_09.dat')
OPEN(12,FILE='pfc_17_03_09.dat')
READ(*,*) INFILE
READ(*,*) OUTFILE
OPEN(14,FILE=INFILE)
OPEN(15,FILE=OUTFILE)
READ(*,*) NSNPLIST
READ(*,*) DRIT
READ(*,*) NWIN
READ(*,*) NREP
READ(*,*) ST
NSG=5770877
C TOTAL NUMBER OF SNPS IN FILE allsnp_gene_NCBI_0KB.txt
NGO=754498
C TOTAL NUMBER OF GENE/GO CATEGORY ASSIGNMENTS
NPFC=28339
C TOTAL NUMBER OF GO CATEGORIES WITH NAMES/FUNCTION DATA
PROCNAM(1)='ALL CATS'
PROCNAM(2)='CELLULAR'
PROCNAM(3)='FUNCTION'
PROCNAM(4)='PROCESS '
PROCNAM(5)='PRO+FUNC'
CRIT(1)=0.05
CRIT(2)=0.01
CRIT(3)=0.001
NUM=0
DO 873 I=1,NPFC
READ(12,*) (IPFC(I,K),K=1,2)
873 CONTINUE
DO 871 I=1,1000000
NGS(I)=0
871 CONTINUE
DO 872 I=1,30000
MGC(I)=0
MCG(I)=0
872 CONTINUE
DO 604 I=10,99
READ(25,*) II,ILIS(I,1),ILIS(I,2)
604 CONTINUE
C READING IN START AND END POINTS ON THE COMPLETE LIST OF SNPS WHOSE
C RS NUMBERS START WITH 10, 11, .... 99
C TO MAKE IT FASTER TO LOCATE THE SNPS USED IN A PARTICULAR STUDY.
MMSP=0
DO 503 I=1,NSG
READ(10,*) ISNP(I),NSNP(I),(JSNP(I,K),K=1,NSNP(I))
503 CONTINUE
C READS IN COMPLETE LIST OF SNPS WITHIN GENES
MGE=0
MSP=0
DO 605 I=1,NSNPLIST
C THIS SECTION READS IN THE SNPS USED IN THE STUDY BEING ANALYSED
C CHECKS THEM AGAINST THE TOTAL LIST OF SNPS WITHIN GENES
C TO CREATE A LIST OF SNPS WITHIN GENES AND THE GENES WITHIN WHICH THEY LIE
READ(14,*) II,PP
RS=REAL(II)
RX=LOG10(RS)
IX=INT(RX)-1
RJ=REAL(IX)
IC=INT(RS/(10**RJ))
IUS=0
DO 696 J=ILIS(IC,1),ILIS(IC,2)
IF (II.EQ.ISNP(J)) THEN
IUS=J
GOTO 697
ENDIF
696 CONTINUE
697 CONTINUE
IF (IUS.EQ.0) THEN
GOTO 605
ENDIF
MSP=MSP+1
IDSNP(MSP)=ISNP(IUS)
NGS(MSP)=NSNP(IUS)
DO 698 J=1,NSNP(IUS)
JGX=JSNP(IUS,J)
JNE=0
DO 802 II=1,MGE
IF (JGX.EQ.IDGE(II)) THEN
JNE=II
ENDIF
802 CONTINUE
IF (JNE.EQ.0) THEN
MGE=MGE+1
IDGE(MGE)=JGX
JNE=MGE
ENDIF
IGS(MSP,J)=JNE
698 CONTINUE
IF (PP.LT.DRIT) THEN
NUM=NUM+1
ISGSNP(NUM)=MSP
ENDIF
C CREATES LIST OF SIGNIFICANT SNPS WITH P-val < SPECIFIED CRITERION
699 CONTINUE
605 CONTINUE
DUM(1)=0.0
DO 603 I=1,MSP
DUM(I+1)=DUM(I)+(1.0/FLOAT(MSP))
603 CONTINUE
DUM(MSP+1)=1.0
PRINT*,'SNPS READ'
PRINT*,'#SNPS = ',MSP
PRINT*,'#GENES = ',MGE
DO 502 I=1,NGO
READ(11,*) JGX,IID
INE=0
DO 842 II=1,MGE
IF (JGX.EQ.IDGE(II)) THEN
INE=II
GOTO 937
ENDIF
842 CONTINUE
937 CONTINUE
IF (INE.EQ.0) THEN
GOTO 502
ENDIF
JNE=0
DO 602 II=1,MMGO
IF (IID.EQ.JDGO(II)) THEN
JNE=II
ENDIF
602 CONTINUE
IF (JNE.EQ.0) THEN
MMGO=MMGO+1
C PRINT*,MMGO,IID
JDGO(MMGO)=IID
JNE=MMGO
ENDIF
INF=0
DO 722 J=1,MGC(INE)
IF (JNE.EQ.JGC(INE,J)) THEN
INF=J
ENDIF
722 CONTINUE
IF (INF.EQ.0) THEN
MGC(INE)=MGC(INE)+1
MCG(JNE)=MCG(JNE)+1
JGC(INE,MGC(INE))=JNE
ENDIF
C JGC CONTAINS ALL GO GATEGORY MEMBERSHIPS FOR EACH GENE
502 CONTINUE
MGO=0
DO 607 I=1,MMGO
C THIS SECTION REMOVES GO CATEGORIES WITH FEWER THAN 3 GENES
C OR MORE THAN 5000 (USER CAN EDIT AND RECOMPILE IF DESIRED)
IF (MCG(I).LT.3.OR.MCG(I).GT.5000) THEN
ICH(I)=0
GOTO 607
ENDIF
MGO=MGO+1
NCG(MGO)=MCG(I)
IDGO(MGO)=JDGO(I)
ICH(I)=MGO
IPF=0
DO 874 J=1,NPFC
IF (IDGO(MGO).EQ.IPFC(J,1)) THEN
IPF=J
GOTO 931
ENDIF
874 CONTINUE
931 CONTINUE
IF (IPF.EQ.0) THEN
PRINT*,'MISSING CATEGORY ',IDGO(MGO)
ELSE
MPFC(MGO)=IPFC(IPF,2)
ENDIF
607 CONTINUE
DO 609 I=1,MGE
NGC(I)=0
NGC(I)=0
DO 608 J=1,MGC(I)
JX=JGC(I,J)
NN=ICH(JX)
IF (NN.NE.0) THEN
NGC(I)=NGC(I)+1
IGC(I,NGC(I))=NN
ENDIF
608 CONTINUE
609 CONTINUE
PRINT*,'CATS READ'
PRINT*,'No.CATS = ',MGO
NMA=0
JMA=0
DO 226 I=1,MGE
IF (NGC(I).GE.NMA) THEN
NMA=NGC(I)
JMA=I
ENDIF
226 CONTINUE
PRINT*,'MAX #CATS PER GENE = ',NMA
PRINT*,'GENE ID ',IDGE(JMA)
NMA=0
JMA=0
DO 2 I=1,MSP
IF (NGS(I).GE.NMA) THEN
NMA=NGS(I)
JMA=I
ENDIF
2 CONTINUE
PRINT*,'NO. SNPS IN STUDY = ',NSNPLIST
PRINT*,'NO. SNPS IN GENES = ',MSP
PRINT*,'MAX #GENES PER SNP = ',NMA
PRINT*,'SNP ID ',IDSNP(JMA)
NMA=0
JMA=0
DO 3 I=1,MGO
IF (NGC(I).GE.NMA) THEN
NMA=NGC(I)
JMA=I
ENDIF
3 CONTINUE
DO 987 IY=1,20000
ISCO(IY)=0
JSCO(IY)=0
987 CONTINUE
DO 587 IY=1,3
DO 787 IZ=1,5
NUMSIG(IY,IZ)=0
787 CONTINUE
587 CONTINUE
JGE=0
DO 14 I=1,NUM
DO 198 J=1,20000
IGOC(J)=0
198 CONTINUE
KU=ISGSNP(I)
C GET THE ID OF EACH SIGNIFICANT SNP
DO 16 J=1,NGS(KU)
ICG=IGS(KU,J)
C WHAT GENES DOES IT LIE IN ?
IUS=0
DO 26 K=1,JGE
IF (ICG.EQ.JSCO(K)) THEN
IUS=K
ENDIF
26 CONTINUE
C IS THIS GENE ALREADY ON THE LIST ?
IF (IUS.EQ.0) THEN
JGE=JGE+1
JSCO(JGE)=ICG
C ADD GENE TO LIST
DO 18 KK=1,NGC(ICG)
K=IGC(ICG,KK)
IF (IGOC(K).EQ.0) THEN
ISCO(K)=ISCO(K)+1
IGOC(K)=1
ENDIF
C INCREASE THE COUNT OF SIGNIFICANT GENES FOR ALL GO CATEGORIES
C CONTAINING THIS GENE BY 1 - BUT ONLY IF THIS SNP HASN'T ALREADY
C CONTRIBUTED A GENE TO THIS CATEGORY. IF A SNP LIES
C IN SEVERAL OVERLAPPING GENES ALL IN THE SAME GO CATEGORY,
C THIS PREVENTS ALL THE GENES BEING COUNTED SEPARATELY.
18 CONTINUE
ENDIF
16 CONTINUE
14 CONTINUE
PRINT*,'P-VALUE CRITERION FOR DEFINING SIGNIFICANT SNPS = ',DRIT
PRINT*,'NUMBER OF GENES WITH SIGNIFICANT SNPS = ',JGE
PRINT*,'NUMBER OF SIMULATED REPLICATE GENE-LISTS = ',NWIN
PRINT*,'NUMBER OF BOOTSTRAP STUDIES = ',NREP
DO 979 IT=1,NWIN
PRINT*,'SIMULATED REPLICATE GENELIST',IT
KRE=1
DO 887 IY=1,10000
JSCO(IY)=0
GM(IY)=0
KSCO(IT,IY)=0
887 CONTINUE
KGE=0
666 CONTINUE
YY=GETRAN(XX,ST)
R=REAL(XX)
KU=0
DO 551 K=1,MSP
IF (R.GE.DUM(K).AND.R.LT.DUM(K+1)) THEN
KU=K
ENDIF
551 CONTINUE
C SELECT A RANDOM SNP
DO 298 J=1,20000
IGOC(J)=0
298 CONTINUE
DO 116 J=1,NGS(KU)
ICG=IGS(KU,J)
C WHAT GENES DOES THE SNP LIE IN ?
IUS=0
DO 216 K=1,KGE
IF (ICG.EQ.JSCO(K)) THEN
IUS=K
ENDIF
216 CONTINUE
C GENE ALREADY ON LIST ?
C PRINT*,II,NID-I,IDJ,IUS,ICG
IF (IUS.EQ.0) THEN
KGE=KGE+1
C PRINT*,IT,KGE
IF (KGE.GT.JGE) THEN
GOTO 994
ENDIF
C GENELIST REACHED THE SAME LENGTH AS THE ORIGINAL ?
JSCO(KGE)=ICG
DO 118 KK=1,NGC(ICG)
K=IGC(ICG,KK)
IF (IGOC(K).EQ.0) THEN
KSCO(IT,K)=KSCO(IT,K)+1
IGOC(K)=1
ENDIF
118 CONTINUE
C ADD 1 TO EACH GO CATEGORY CONTAINING THE GENE (SEE ABOVE)
ENDIF
116 CONTINUE
GOTO 666
994 CONTINUE
979 CONTINUE
DO 812 IX=1,5
PMIN(IX)=99.0
812 CONTINUE
NSCA=0
DO 22 I=1,MGO
NMP=MPFC(I)
IF (ISCO(I).LE.1) THEN
GOTO 22
ENDIF
C ONLY COUNT GO CATEGORIES WITH TWO OR MORE SIGNIFICANT GENES
PR(I)=0
ZR(I)=0
RU(I)=0.0d0
CUM(1)=0
DO 32 J=1,NWIN
IF (ISCO(I).LE.KSCO(J,I)) THEN
PR(I)=PR(I)+(1.0/FLOAT(NWIN))
ENDIF
C CATEGORY-SPECIFIC P-VALUE = NUMBER OF SIMULATED GENELISTS
C WHERE THE SCORE FOR THE CATEGORY (KSCO) IS AT LEAST AS HIGH
C AS THAT IN THE REAL DATA (ISCO)
CUM(J+1)=CUM(J)+(1.0/FLOAT(NWIN))
GM(I)=GM(I)+(FLOAT(KSCO(J,I))/FLOAT(NWIN))
C CALCULATING EXPECTED NUMBER OF SIGNIFICANT GENES IN THE CATEGORY
32 CONTINUE
CUM(NWIN+1)=1.0
IF (PR(I).LT.PMIN(1)) THEN
PMIN(1)=PR(I)
ENDIF
DO 813 J=1,3
IF (NMP.EQ.J.AND.PR(I).LE.PMIN(J+1)) THEN
PMIN(J+1)=PR(I)
ENDIF
813 CONTINUE
IF (NMP.GE.2.AND.PR(I).LE.PMIN(5)) THEN
PMIN(5)=PR(I)
ENDIF
C CALCULATING MINIMUM P-VALUES FOR CATEGORIES OF EACH FUNCTIONAL TYPE
DO 33 J=1,3
IF (PR(I).LE.CRIT(J)) THEN
NUMSIG(J,1)=NUMSIG(J,1)+1
NUMSIG(J,NMP+1)=NUMSIG(J,NMP+1)+1
IF (NMP.GE.2) THEN
NUMSIG(J,5)=NUMSIG(J,5)+1
ENDIF
ENDIF
C COUNTING NUMBER OF CATEGORIES REACHING A GIVEN CATEGORY-SPECIFIC P-VALUE
33 CONTINUE
QP(I)=DBLE(PR(I))
22 CONTINUE
NBOOT=NWIN
DO 716 IC=1,3
DO 816 ID=1,5
YR(IC,ID)=0
816 CONTINUE
716 CONTINUE
DO 819 I=1,5
QZ(I)=0
819 CONTINUE
DO 51 I=1,NREP
C PRINT*,I
C PERFORMING REPLICATE "STUDIES" TO GET SIGNIFICANCE OF NUMBERS OF
C SIGNIFICANTLY OVER-REPRESENTED CATEGORIES
YY=GETRAN(XX,ST)
R=REAL(XX)
IYS=0
DO 53 K=1,NWIN
IF (R.GE.CUM(K).AND.R.LT.CUM(K+1)) THEN
IYS=K
ENDIF
53 CONTINUE
C CHOOSE A REPLICATE GENE-LIST TO ACT AS THE "REAL DATA"
DO 54 K=1,MGO
MSCO(K)=KSCO(IYS,K)
QR(K)=0
54 CONTINUE
DO 58 K=1,3
DO 589 KK=1,5
MUMSIG(K,KK)=0
589 CONTINUE
58 CONTINUE
DO 52 J=1,NBOOT
528 CONTINUE
YY=GETRAN(XX,ST)
R=REAL(XX)
JYS=0
DO 55 K=1,NWIN
IF (R.GE.CUM(K).AND.R.LT.CUM(K+1)) THEN
JYS=K
ENDIF
55 CONTINUE
IF (JYS.EQ.IYS) THEN
GOTO 528
ENDIF
C SELECT REPLICATE GENE LISTS (WITH REPLACEMENT)
DO 56 K=1,MGO
IF (MSCO(K).LE.KSCO(JYS,K)) THEN
QR(K)=QR(K)+(1.0/FLOAT(NBOOT))
ENDIF
C OBTAIN CATEGORY-SPECIFIC P-VALUES FOR THE "REAL DATA"
56 CONTINUE
52 CONTINUE
DO 814 K=1,5
QMIN(K)=99.0
814 CONTINUE
DO 57 K=1,MGO
NMP=MPFC(K)
IF (MSCO(K).LE.1) THEN
GOTO 57
ENDIF
JGG=IDGO(K)
IF (QR(K).LT.QMIN(1)) THEN
QMIN(1)=QR(K)
ENDIF
DO 815 JX=1,3
IF (NMP.EQ.JX.AND.QR(K).LE.QMIN(JX+1)) THEN
QMIN(JX+1)=QR(K)
ENDIF
815 CONTINUE
IF (NMP.GE.2.AND.QR(K).LE.QMIN(5)) THEN
QMIN(5)=QR(K)
ENDIF
C OBTAIN MINIMUM P-VALUES FOR CATEGORIES OF VARIOUS FUNCTIONAL TYPES
DO 63 JJ=1,3
IF (QR(K).LE.CRIT(JJ)) THEN
MUMSIG(JJ,1)=MUMSIG(JJ,1)+1
MUMSIG(JJ,NMP+1)=MUMSIG(JJ,NMP+1)+1
IF (NMP.GE.2) THEN
MUMSIG(JJ,5)=MUMSIG(JJ,5)+1
ENDIF
ENDIF
C OBTAIN NUMBERS OF CATEGORIES WITH CATEGORY-SPECIFIC P-VALUES LESS THAN
C VARIOUS CRITERIA
63 CONTINUE
57 CONTINUE
DO 67 K=1,MGO
IF (ISCO(K).LE.1) THEN
GOTO 67
ENDIF
IF (PR(K).GE.QMIN(1)) THEN
ZR(K)=ZR(K)+(1.0/FLOAT(NREP))
ENDIF
C PROBABILITY THAT ANY CATEGORY-SPECIFIC P-VALUE IN A STUDY WHERE THERE
C IS NO FUNCTIONAL ASSOCIATION (i.e. RANDOMLY-SELECTED "REAL DATA") IS
C LESS THAN OR EQUAL TO THE P-VALUE OF A GIVEN CATEGORY IN THE ORIGINAL
C DATASET (i.e. A STUDY-WIDE P-VALUE CORRECTED FOR TESTING MULTIPLE NON-
C INDEPENDENT CATEGORIES)
DO 77 KK=1,MGO
IF (MSCO(KK).LE.1) THEN
GOTO 77
ENDIF
IF (PR(K).GE.QR(KK)) THEN
RU(K)=RU(K)+(1.0d0/DFLOAT(NREP))
ENDIF
C EXPECTED NUMBER OF CATEGORIES PER STUDY WITH CATEGORY-SPECIFIC P-vALUES
C LESS THAN OR EQUAL TO THAT OF A GIVEN CATEGORY IN THE ORIGINAL DATASET
77 CONTINUE
67 CONTINUE
DO 817 JJ=1,5
IF (QMIN(JJ).LE.PMIN(JJ)) THEN
QZ(JJ)=QZ(JJ)+(1.0/FLOAT(NREP))
ENDIF
C PROBABILITY THAT THE MINIMUM CATEGORY-SPECIFIC P-VALUE IN A STUDY WHERE THERE
C IS NO FUNCTIONAL ASSOCIATION (i.e. RANDOMLY-SELECTED "REAL DATA") IS
C LESS THAN OR EQUAL TO THE MINIMUM CATEGORY-SPECIFIC P-VALUE IN THE ORIGINAL
C DATA
817 CONTINUE
DO 64 JJ=1,3
DO 642 JH=1,5
IF (MUMSIG(JJ,JH).GE.NUMSIG(JJ,JH)) THEN
YR(JJ,JH)=YR(JJ,JH)+(1.0/FLOAT(NREP))
ENDIF
C PROBABILITY THAN THE NUMBER OF CATEGORIES REACHING SIGNIFICANCE IN A STUDY
C WHERE THERE IS NO FUNCTIONAL ASSOCIATION (i.e. RANDOMLY-SELECTED "REAL DATA")
C IS LESS THAN OR EQUAL TO THAT IN THE ORIGINAL DATA (i.e. TESTING FOR A
C SIGNIFICANTLY LARGE NUMBER OF OVER-REPRESENTED CATEGORIES)
642 CONTINUE
64 CONTINUE
51 CONTINUE
DO 712 IT=1,MGO
IF (ISCO(IT).EQ.0) THEN
QP(IT)=1d0+DBLE(GM(IT))
ENDIF
712 CONTINUE
X=QUIKSORT(QP,MGO,MUM)
DO 23 I=1,MGO
J=MUM(I)
NMP=MPFC(J)
PXN=PROCNAM(NMP+1)
JG=IDGO(J)
JH=ISCO(J)
IF (ISCO(J).LE.1) THEN
GOTO 23
ENDIF
C ONLY COUNT GO CATEGORIES WITH TWO OR MORE SIGNIFICANT GENES
WRITE(15,44) JG,PXN,JH,JGE,NCG(J),MGE,GM(J),PR(J),ZR(J),RU(J)
23 CONTINUE
WRITE(15,*)
WRITE(15,*)
DO 691 ISS=1,5
WRITE(15,*) PROCNAM(ISS)
WRITE(15,*)
WRITE(15,*) 'MIN P =',PMIN(ISS),' SIG = ',QZ(ISS)
WRITE(15,*)
DO 69 IS=1,3
WRITE(15,*) 'NO. OF P<',CRIT(IS),' =',NUMSIG(IS,ISS)
WRITE(15,*) 'SIG = ',YR(IS,ISS)
69 CONTINUE
WRITE(15,*)
691 CONTINUE
CLOSE(15)
44 FORMAT(I8,2X,A8,3(2X,I5),2X,I5,2X,F7.2,2X,F7.5,2X,F6.4,2X,F9.2)
STOP
END
REAL*8 FUNCTION GETRAN(X,S)
REAL*8 X,S,A,B,SN,SS,RN
INTEGER IS
A=7D0**5
B=(2D0**31)-1
SN=S*A
IS=INT(SN/B)
SS=DFLOAT(IS)
C PRINT*,SN,SS,IS
RN=SN-(SS*B)
S=RN
X=S/B
GETRAN=1D0
RETURN
END
REAL FUNCTION QUIKSORT(RLOD,NLO,NUM)
REAL*8 RLOD(20000)
REAL*8 SORT1(20000),SORT2(20000),SORT3(20000)
INTEGER NUM(20000),NUM1(20000),NUM2(20000)
DO 5 I=1,NLO
SORT1(I)=-1000.0D0
NUM1(I)=0
5 CONTINUE
DO 10 I=1,NLO
XLO=RLOD(I)
C PRINT*,XLO
DO 20 J=NLO+1-I,NLO
IF (XLO.GE.SORT1(J)) THEN
C PRINT*,XLO,SORT1(J),J
IPOS=J
ENDIF
20 CONTINUE
C PRINT*,IPOS
SORT2(IPOS)=XLO
NUM2(IPOS)=I
DO 30 K=2,IPOS
SORT2(K-1)=SORT1(K)
NUM2(K-1)=NUM1(K)
30 CONTINUE
DO 40 K=IPOS+1,NLO
SORT2(K)=SORT1(K)
NUM2(K)=NUM1(K)
40 CONTINUE
DO 50 L=1,NLO
NUM1(L)=NUM2(L)
SORT1(L)=SORT2(L)
C PRINT*,L,SORT1(L)
50 CONTINUE
10 CONTINUE
DO 60 I=1,NLO
NUM(I)=NUM1(I)
60 CONTINUE
QUIKSORT=1.0
RETURN
END
因为我不是fortran-coder，不知道这段代码能不能改成完全支持64位系统？
编辑： 我试过这样编译：
gfortran aligator.f -o aligator -g -fcheck=all -fbacktrace
有这些错误：
/tmp/ccbJsCPS.o: In function `MAIN__':
/home/alexsson/Programs/Aligator/aligator.f:354:(.text+0x628a): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:355:(.text+0x62ff): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:356:(.text+0x6374): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:357:(.text+0x6380): relocation truncated to fit: R_X86_64_PC32 against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:360:(.text+0x65c5): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:360:(.text+0x65eb): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:365:(.text+0x66c3): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:365:(.text+0x66e9): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:369:(.text+0x697d): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:370:(.text+0x69ee): relocation truncated to fit: R_X86_64_32S against `.bss'
/home/alexsson/Programs/Aligator/aligator.f:371:(.text+0x6a6a): additional relocation overflows omitted from the output
collect2: error: ld returned 1 exit status
我不确定我是否变得更聪明了...
解法： 感谢 tkoenig，解决方案是将 aligator.f 的源代码更改为：http://pastebin.com/tnYhuba3
它适用于我的 Slackware 64 位当前版本（已启用 multilib）- 具有 16 GB RAM 的系统。
你有一些真的 大数组，这会导致 32 位寻址模式出现问题。如果您使数组可分配并在程序开始时分配它们，就像 with
CHARACTER*50 INFILE,OUTFILE
CHARACTER*8 PROCNAM(5),PXN
integer, dimension(:,:), allocatable :: jsnp
integer, dimension(:), allocatable :: nsnp, isnp
INTEGER ILIS(100,2)
INTEGER NGS(1000000),IDSNP(1000000)
INTEGER IGS(1000000,25)
INTEGER ICH(20000),IGOC(20000)
INTEGER IDGE(30000),IPFC(35000,2)
INTEGER IDGO(20000),NGC(30000)
INTEGER JDGO(20000),MGC(30000)
INTEGER MPFC(20000),MCG(30000),NCG(30000)
integer, dimension(:,:), allocatable :: igc, jgc, ksco
INTEGER JSCO(20000)
INTEGER ISGSNP(200000)
INTEGER MSCO(20000)
INTEGER NUMSIG(3,5)
INTEGER MUMSIG(3,5),ISLIM(200)
INTEGER ISCO(20000),MUM(20000),NUN(20000)
REAL PR(20000),CRIT(3),CUM(1000001),QR(20000)
REAL ZR(20000),YR(3,5),GM(20000),QZ(5),PMIN(5)
REAL DUM(1000001),QMIN(5)
REAL*8 QP(20000),CHX,RU(20000),QQ(20000)
REAL*8 XX,ST,YY
REAL*8 GETRAN
EXTERNAL GETRAN,QUIKSORT
allocate(JSNP(6000000,25))
allocate(NSNP(6000000), ISNP(6000000))
allocate(IGC(30000,1000),JGC(30000,1000),KSCO(50000,10000))
这应该不再是一个问题。
内存不足，无法查看。
根据您需要为 INTEGER 使用的值的范围，将默认整数更改为两字节或一字节的类型可能也很有趣，如
integer, parameter :: i2 = selected_int_kind(4) ! select range of at least 10000, usually translates into two bytes
integer(kind=i2), dimension(:,:), allocatable :: jsnp
或
integer, parameter :: i1 = selected_int_kind(2) ! one-byte integer
这可以显着节省内存和速度。