ancestry: allow globs

closes #59

CHANGES.md

@@ -2,6 +2,7 @@ v0.2.12 (dev)
 =============
 + add checkbox to HTML to scale IBS0, IBS2, etc by number of sites shared by the samples. this almost always
   results in a scaling that is better across (pairs of) samples.
++ ancestry: allow globs for ancestry files (#59)
 
 
 v0.2.11

src/somalierpkg/ancestry.nim

@@ -9,6 +9,7 @@ import strformat
 import arraymancer
 import ./relate
 import ./depthview
+import ./common
 import argparse
 import sets
 import arraymancer
@@ -104,6 +105,8 @@ proc ancestry_main*() =
     echo p.help
     quit "send argument for extracted files"
 
+  opts.extracted.update_with_glob
+
   var (labeled_samples, query_samples) = opts.extracted.split_labeled_samples
 
   var train_mat = newSeq[seq[float32]](labeled_samples.len)

src/somalierpkg/common.nim

@@ -3,6 +3,7 @@ import strformat
 import algorithm
 import hts/private/hts_concat
 import hts/fai
+import os
 
 const formatVersion* = 2'u8
 
@@ -16,6 +17,21 @@ type Site* = object
   ## counts if swap is true.
   flip*:bool
 
+
+proc update_with_glob*(files: var seq[string]) =
+  var toadd = newSeqOfCap[string](256)
+  for i in 0..<min(files.len, 10):
+    if files[i] == "++":
+      toadd.add(files[i])
+      continue
+    for w in files[i].walkFiles:
+      toadd.add(w)
+
+  if files.len > 10:
+    files = toadd & files[10..files.high]
+  else:
+    files = toadd
+
 {.push checks: off, optimization:speed.}
 proc toSite(toks: seq[string]): Site =
   result = Site()

src/somalierpkg/relate.nim

@@ -868,17 +868,6 @@ proc toj(sample_names: seq[string], stats: seq[Stat4], gt_counts: array[5, seq[u
     ))
   result.add("]")
 
-proc update_with_glob(files: var seq[string]) =
-  var toadd = newSeqOfCap[string](256)
-  for i in 0..<min(files.len, 10):
-    for w in files[i].walkFiles:
-      toadd.add(w)
-
-  if files.len > 10:
-    files = toadd & files[10..files.high]
-  else:
-    files = toadd
-
 proc add_prefixed_samples(groups: var seq[pair], samples: seq[string], prefixes: seq[string]) =
   # update groups so that sample == ${prefix}sample
   #if len(prefixes) == 0: return
@@ -1034,7 +1023,7 @@ specified as comma-separated groups per line e.g.:
       relGt0p2.mgetOrPut(rel.sample_a, @[]).add(rel.sample_b)
       relGt0p2.mgetOrPut(rel.sample_b, @[]).add(rel.sample_a)
 
-    let ra = random(1'f32)
+    let ra = rand(1'f32)
     let interesting = expected_relatedness != -1 or rr > 0.05
     if (ra > proportion_sampled) and not interesting:
       continue