Merge pull request #16 from 4dn-dcic/dev

Dev

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: Rpairix
 Title: Rpairix
-Version: 0.1.0
+Version: 0.1.1
 Authors@R: person("Soo", "Lee", email = "duplexa@gmail.com", role = c("aut", "cre"))
 Description: R binder for pairix, tool for querying a pair of genomic ranges in a pairs file (pairix-indexed bgzipped text file)
 Depends:

R/px_build_index.R

@@ -40,6 +40,7 @@ px_build_index<-function(filename, preset='', sc=0, bc=0, ec=0, sc2=0, bc2=0, ec
   if(out[[13]][1] == -2) { message("Can't recognize preset."); return(-1); }
   if(out[[13]][1] == -3) { message("Was bgzip used to compress this file?"); return(-1); }
   if(out[[13]][1] == -4) { message("The index file exists. Please use force=TRUE to overwrite"); return(-1); }
+  if(out[[13]][1] == -5) { message("Can't recognize file type, with no preset specified."); return(-1); }
   return(0);
 }
 

R/px_query.R

@@ -15,7 +15,7 @@
 #' 
 #' ## 2D-indexed file
 #' filename = system.file(".","test_4dn.pairs.gz", package="Rpairix")
-#' querystr = c("chr10|chr20","chr2|chr2")
+#' querystr = c("chr10|chr20","chr22|chr22")
 #' res = px_query(filename, querystr)
 #' print(res)
 #'
@@ -48,7 +48,7 @@
 #'
 #' ## 1D-indexed file
 #' filename = system.file(".","SRR1171591.variants.snp.vqsr.p.vcf.gz", package="Rpairix")
-#' querystr = 'chr10'
+#' querystr = 'chr10|5000000-20000000'
 #' res = px_query(filename, querystr)
 #' print(res)
 #' 

README.md

@@ -34,7 +34,7 @@ R --no-site-file --no-environ --no-save --no-restore CMD INSTALL --install-tests
 To install a specific version,
 ```
 library(devtools)
-install_url("https://github.com/4dn-dcic/Rpairix/archive/0.1.0.zip")
+install_url("https://github.com/4dn-dcic/Rpairix/archive/0.1.1.zip")
 ```
 
 
@@ -78,6 +78,7 @@ sc2 second sequence (chromosome) column index (1-based). Zero (0) means not spec
 * `line_skip` : number of lines to skip in the beginning. (default 0)
 * `force` : If TRUE, overwrite existing index file. If FALSE, do not overwrite unless the index file is older than the bgzipped file. (default FALSE)
 * An index file sometextfile.gz.px2 will be created.
+* When neither `preset` nor `sc`(and `bc`) is given, the following file extensions are automatically recognized: `gff.gz`, `bed.gz`, `sam.gz`, `vcf.gz`, `psltbl.gz` (1D-indexing), and `pairs.gz` (2D-indexing).
 
 ### Querying
 ```
@@ -155,6 +156,7 @@ px_check_dim(filename)
 > # indexing
 > px_build_index(filename, sc=2, bc=3, ec=3, sc2=4, bc2=5, ec2=5, force=TRUE)
 > px_build_index(filename, 'pairs', force=TRUE)  # equivalent to the above line
+> px_build_index(filename, force=TRUE)  # equivalent to the above line, since file extension pairs.gz is recognized.
 >
 > # single-query
 > querystr = "chr10:1-3000000|chr20"
@@ -239,6 +241,9 @@ Individual R functions are written and documented in `R/`. The `src/rpairixlib.c
 
 
 ## Version history
+### 0.1.1
+* `px_build_index`: When neither `preset` nor a custom set of columns is given, file extensions are automatically recognized for indexing.
+
 ### 0.1.0
 * `px_build_index` is added. (Now indexing can be done using Rpairix as well as querying.)
 

Rpairix.Rproj

@@ -1,4 +1,4 @@
-Version: 0.1.0
+Version: 0.1.1
 
 RestoreWorkspace: No
 SaveWorkspace: No

src/rpairixlib.c

@@ -355,7 +355,18 @@ void build_index(char **pinputfilename, char **ppreset, int *psc, int *pbc, int
     if ( bgzf_is_bgzf(*pinputfilename)!=1 ) *pflag = -3;
     else {
       ti_conf_t conf;
-      if (strcmp(*ppreset, "") == 0 && *psc != 0 && *pbc != 0){
+      if (strcmp(*ppreset, "") == 0 && *psc == 0 && *pbc == 0){
+        int l = strlen(*pinputfilename);
+        int strcasecmp(const char *s1, const char *s2);
+        if (l>=7 && strcasecmp(*pinputfilename+l-7, ".gff.gz") == 0) conf = ti_conf_gff;
+        else if (l>=7 && strcasecmp(*pinputfilename+l-7, ".bed.gz") == 0) conf = ti_conf_bed;
+        else if (l>=7 && strcasecmp(*pinputfilename+l-7, ".sam.gz") == 0) conf = ti_conf_sam;
+        else if (l>=7 && strcasecmp(*pinputfilename+l-7, ".vcf.gz") == 0) conf = ti_conf_vcf;
+        else if (l>=10 && strcasecmp(*pinputfilename+l-10, ".psltbl.gz") == 0) conf = ti_conf_psltbl;
+        else if (l>=9 && strcasecmp(*pinputfilename+l-9, ".pairs.gz") == 0) conf = ti_conf_pairs;
+        else *pflag = -5; // file extension not recognized and no preset specified
+      }
+      else if (strcmp(*ppreset, "") == 0 && *psc != 0 && *pbc != 0){
         conf.sc = *psc;
         conf.bc = *pbc;
         conf.ec = *pec;
@@ -376,7 +387,7 @@ void build_index(char **pinputfilename, char **ppreset, int *psc, int *pbc, int
       else if (strcmp(*ppreset, "old_merged_nodups") == 0) conf = ti_conf_old_merged_nodups;
       else *pflag = -2;  // wrong preset
 
-      if (*pflag != -2 ) *pflag= ti_index_build(*pinputfilename, &conf);  // -1 if failed
+      if (*pflag != -2 && *pflag != -5 ) *pflag= ti_index_build(*pinputfilename, &conf);  // -1 if failed
     }
   }
 }