#! /bin/tcsh -f
#
#
#

# extract REMARK 200 from everything
find ftp.wwpdb.org/pub/pdb/data/structures/divided/pdb/ -name '*.ent.gz' -print -exec zgrep "REMARK 200" \{\} \; |\
awk '/^ftp./{id=substr($0,length($0)-10,4)}\
 /^REMARK 200 /{rem=1;print id,$0}\
 id!=lastid && lastid!="" && rem==0{print lastid,"has no REM200"}\
 id!=lastid {rem=0}\
 {lastid=id}' |\
cat >! REM200.txt

# get all information that can be used to determine the method used
cat REM200.txt |\
 awk -F ":" '\
    /METHOD USED TO DETERMINE THE STRUCTURE/{\
     id=substr($0,1,4);meth=$NF;\
     getline;\
     while(substr($0,17,13)!="SOFTWARE USED" && substr($0,17)~/[^ ]/){meth=meth substr($0,17);getline}\
     while(gsub("  "," ",meth));\
     while(gsub(":"," ",meth));}\
     /STARTING MODEL/ && ! /NULL/{meth="MOLECULAR REPLACEMENT " meth}\
     /STARTING MODEL/{print id,":",type,":",meth}\
     /EXPERIMENT TYPE/{type=$NF}' |\
cat > ! method.txt


# reformat it a bit
awk -F ":" '{print $1,"isapdb"}' method.txt |\
cat >! stuff
awk -F ":" '{print $NF}' method.txt |\
cat >> stuff

# try to figure out what it all means
cat stuff |\
awk '$NF=="isapdb"{++isapdb[toupper($1)];next}\
     /MOLECULAR/ || /REPLACEMENT/ {++count["MR"];next}\
     /STRUCTURE/ {++count["MR"];next}\
     /NATIVE/ {++count["MR"];next}\
     / PDB / {++count["MR"];next}\
     / STARTING MODEL / {++count["MR"];next}\
     / WILD-TYPE / {++count["MR"];next}\
     /COORDINATES/ {++count["MR"];next}\
     /DIRECT REFINE/ {++count["MR"];next}\
     /RIGID.BODY/ {++count["MR"];next}\
     /ALREADY SOLVED/ {++count["MR"];next}\
     /[- ]MAD[- ]/ {++count["MAD"];next}\
     /[- ]RIP[- ]/ {++count["RIP"];next}\
     /SEMAD/ {++count["MAD"];next}\
     /HGMAD/ {++count["MAD"];next}\
     /SINGLE/ && /ANOM[AO]LOUS/ {++count["SAD"];next}\
     /ANOM[AO]LOUS/ {++count["MAD"];next}\
     /[- ]SAD[- ]/ {++count["SAD"];next}\
     /[- ]SAS[- ]/ {++count["SAD"];next}\
     /[- ]SIR[- ]/ {++count["SIR"];next}\
     / SINGLE WAVELENGTH / {++count["SAD"];next}\
     / ISOMORPHOUS REPLACEMENT / {++count["MIR"];next}\
     / MIR / {++count["MIR"];next}\
     / M\.I\.R/ {++count["MIR"];next}\
     / MULTIPLE ISOMORPHOUS / {++count["MIR"];next}\
     / SINGLE ISOMORPHOUS / {++count["SIR"];next}\
     / SIRA[SD]/ {++count["SIRAS"];next}\
     / SIR[SO\/]A[SD]/ {++count["SIRAS"];next}\
     / SIR[\(]AS/ {++count["SIRAS"];next}\
     / MIRA[SD]/ {++count["MIRAS"];next}\
     / MIR[SO\/]A[SD]/ {++count["MIRAS"];next}\
     / MIR[\(]AS/ {++count["MIRAS"];next}\
     / HEAVY/ {++count["MIR"];next}\
     / SOAKING/ {++count["MIR"];next}\
     / SIMULA/{++count["MR"];next}\
     /PDB CODE/{++count["MR"];next}\
     / WILD/ && /TYPE /{++count["MR"];next}\
     / SINGLE HG/ || / ISAS /{++count["SIR"];next}\
     /AUTOSOL /{++count["MAD"];next}\
     / REFMAC/{++count["MR"];next}\
     / MOLREP/{++count["MR"];next}\
     / PHASER $/{++count["MR"];next}\
     / PHENIX /{++count["MR"];next}\
     / COMO /{++count["MR"];next}\
     / SOLVE $/ || /AUTOSHARP/{++count["MAD"];next}\
     / ARP\/WARP / {++count["MR"];next}\
     / RIPAS/ {++count["RIP"];next}\
     /SAD/ {++count["SAD"];next}\
     /MAD/ {++count["MAD"];next}\
     /MIR/ {++count["MIR"];next}\
     /SIR/ {++count["SIR"];next}\
     /MR/ {++count["MR"];next}\
     isapdb[$NF]{++count["MR"];next}\
     /REFINEMENT/ {++count["MR"];next}\
     /D-FOURI/ || /DIFFERENCE/ {++count["MR"];next}\
     /DIRECT/ || /HALF-BAKE/ || /AB INITIO/ || /DUAL-SPACE/ {++count["AI"];next}\
     / WAVELENGTH / {++count["MAD"];next}\
     / ISMORPHOUS/ || /ISOMORPH/ && ! /REPLA/ || /ISO TO/ {++count["MR"];next}\
     / FOURIER SYNTHESIS / || / FOURIER METHODS / {++count["MR"];next}\
     / KNOWN / {++count["MR"];next}\
     / MANUALLY PLACED / {++count["MR"];next}\
     {++count[$0]} END{\
     for(meth in count) print count[meth],meth}' |\
 sort -g | tee method_count.txt