Merge pull request #27 from karel-brinda/output

Improve cli, output, and tests
karel-brinda · Apr 9, 2024 · 1004637 · 1004637
2 parents ded070c + bd940a4
commit 1004637
Show file tree

Hide file tree

Showing 6 changed files with 94 additions and 45 deletions.
diff --git a/README.rst b/README.rst
@@ -67,30 +67,29 @@ Command-line parameters
 .. code-block::
 
     $ attotree -h
-
+    
     Program: attotree (rapid estimation of phylogenetic trees using sketching)
-    Version: 0.1.5
+    Version: 0.1.6
     Author:  Karel Brinda <karel.brinda@inria.fr>
 
-    usage: attotree [-k INT] [-s INT] [-t INT] [-o FILE] [-f STR] [-L] [-D] [-V] genomes [genomes ...]
+    usage: attotree [-k INT] [-s INT] [-t INT] [-o FILE] [-m STR] [-d DIR] [-L] [-D] [-V] genome [genome ...]
 
     positional arguments:
-      genomes     input genome file (fasta / gzipped fasta / list of files when "-L")
+      genome      input genome file(s) (fasta / gzipped fasta / list of files when "-L")
 
     options:
       -h          show this help message and exit
       -v          show program's version number and exit
       -k INT      kmer size [21]
       -s INT      sketch size [10000]
-      -t INT      number of threads [10]
-      -o FILE     newick output [stdout]
-      -f STR      tree inference algorithm (nj/upgma) [nj]
+      -t INT      number of threads [#cores, 10]
+      -o FILE     newick output [-]
+      -m STR      tree construction method (nj/upgma) [nj]
+      -d DIR      tmp dir [default system, /var/folders/z6...]
       -L          input files are list of files
       -D          debugging (don't remove tmp dir)
       -V          verbose output
 
-
-
 Issues
 ------
 

diff --git a/attotree/attotree.py b/attotree/attotree.py
@@ -26,7 +26,7 @@
 DEFAULT_S = 10000
 DEFAULT_K = 21
 DEFAULT_T = os.cpu_count()
-DEFAULT_F = "nj"
+DEFAULT_M = "nj"
 
 
 def shorten_output(s):
@@ -267,13 +267,13 @@ def postprocess_quicktree_nw(nw_in_fn, nw_out_fn, verbose):
             buffer.append(x)
     s = "".join(buffer)
     if nw_out_fn == "-":
-        print(s)
+        print(s, end="")
     else:
         with open(nw_out_fn, "w+") as fo:
-            print(s, file=fo)
+            fo.write(s)
 
 
-def attotree(fns, newick_fn, k, s, t, phylogeny_algorithm, fof, verbose, debug):
+def attotree(fns, newick_fn, k, s, t, phylogeny_algorithm, tmp_dir, fof, verbose, debug):
     """
     Generate a phylogenetic tree using the given parameters.
 
@@ -284,6 +284,7 @@ def attotree(fns, newick_fn, k, s, t, phylogeny_algorithm, fof, verbose, debug):
         s (int): Value for parameter s.
         t (int): Value for parameter t.
         phylogeny_algorithm (str): Name of the phylogeny algorithm to use.
+        tmp_dir (str): Temporary directory.
         fof (bool): Flag indicating whether to use the fof parameter.
         verbose (bool): Flag indicating whether to enable verbose output.
         debug (bool): Flag indicating whether to retain auxiliary files.
@@ -302,7 +303,9 @@ def attotree(fns, newick_fn, k, s, t, phylogeny_algorithm, fof, verbose, debug):
         fmsg = ""
     message(f"Attotree starting{fmsg}")
 
-    d = tempfile.mkdtemp()
+    message(str(tmp_dir))
+    d = tempfile.mkdtemp(dir=tmp_dir)
+    message(d)
 
     message('Creating a temporary directory', d)
     phylip1_fn = os.path.join(d, "distances.phylip0")
@@ -311,12 +314,13 @@ def attotree(fns, newick_fn, k, s, t, phylogeny_algorithm, fof, verbose, debug):
     newick2_fn = newick_fn
     if fof:
         #This is to make the list of file pass to Mash even with
-        #process substitutions
-        old_fof_fn = fns[0]
+        #process substitutions and allows for merging mutliple lists
         new_fof_fn = os.path.join(d, "fof.txt")
-        with open(old_fof_fn) as f, open(new_fof_fn, 'w') as g:
-            g.write(f.read())
-        fns = [new_fof_fn]
+        with open(new_fof_fn, 'w') as g:
+            for old_fof_fn in fns:
+                with open(old_fof_fn) as f:
+                    g.write(f.read().strip() + "\n")
+                fns = [new_fof_fn]
     mash_triangle(fns, phylip1_fn, k=k, s=s, t=t, fof=fof, verbose=verbose)
     postprocess_mash_phylip(phylip1_fn, phylip2_fn, verbose=verbose)
     quicktree(phylip2_fn, newick1_fn, algorithm=phylogeny_algorithm, verbose=verbose)
@@ -408,7 +412,7 @@ def format_help(self):
         metavar='INT',
         dest='t',
         default=DEFAULT_T,
-        help=f'number of threads [{DEFAULT_T}]',
+        help=f'number of threads [#cores, {DEFAULT_T}]',
     )
 
     parser.add_argument(
@@ -420,19 +424,27 @@ def format_help(self):
     )
 
     parser.add_argument(
-        '-f',
+        '-m',
         metavar='STR',
-        dest='f',
-        default=DEFAULT_F,
+        dest='m',
+        default=DEFAULT_M,
         choices=("nj", "upgma"),
-        help=f'tree inference algorithm (nj/upgma) [{DEFAULT_F}]',
+        help=f'tree construction method (nj/upgma) [{DEFAULT_M}]',
+    )
+
+    parser.add_argument(
+        '-d',
+        metavar='DIR',
+        dest='d',
+        default=None,
+        help=f'tmp dir [default system, {tempfile.gettempdir()[:15]+"..."}]',
     )
 
     parser.add_argument(
         '-L',
         action='store_true',
         dest='L',
-        help=f'input files are list of files',
+        help=f'input files are list(s) of files',
     )
 
     parser.add_argument(
@@ -450,17 +462,17 @@ def format_help(self):
     )
 
     parser.add_argument(
-        'genomes',
+        'genome',
         nargs="+",
-        help='input genome file (fasta / gzipped fasta / list of files when "-L")',
+        help='input genome file(s) (fasta / gzipped fasta / list of files when "-L")',
     )
 
     args = parser.parse_args()
 
     #print(args)
     attotree(
-        fns=args.genomes, k=args.k, s=args.s, t=args.t, newick_fn=args.o, phylogeny_algorithm=args.f, fof=args.L,
-        verbose=args.V, debug=args.D
+        fns=args.genome, k=args.k, s=args.s, t=args.t, newick_fn=args.o, phylogeny_algorithm=args.m, fof=args.L,
+        verbose=args.V, debug=args.D, tmp_dir=args.d
     )
 
     args = parser.parse_args()

diff --git a/attotree/version.py b/attotree/version.py
@@ -2,4 +2,4 @@
     from __commit import *
 except ImportError:
     pass
-VERSION = "0.1.5"
+VERSION = "0.1.6"
diff --git a/tests/02_simple_tree/Makefile b/tests/02_simple_tree/Makefile
@@ -1,4 +1,4 @@
-.PHONY: all help clean cleanall view
+.PHONY: all all1 all2 help clean cleanall view data
 
 SHELL=/usr/bin/env bash -eo pipefail
 
@@ -8,53 +8,87 @@ SHELL=/usr/bin/env bash -eo pipefail
 
 FASTAS=$(shell tar -tf test_spneumo.tar.xz)
 
-###################
-## Compare trees ##
-###################
+##########
+## Main ##
+##########
+all: ## Run all tests
+all: all1 all2
 
-all: \
-	tree.precomputed.newick \
+#w/diff
+all1: \
 	tree.default.nw \
+	tree.default_local_tmp_dir.nw \
 	tree.default_verbose_debug.nw \
 	tree.file_list.nw \
-	tree.file_list_proc_subst.nw
+	tree.file_list_proc_subst.nw \
+	tree.file_list_proc_subst_two_lists.nw \
+	tree.default_o-param.nw
 	for x in $^; do \
 		2>&1 echo Testing "$$x"; \
 		diff $$x tree.precomputed.newick; \
 	done
 
-###################################
-# Different ways to compute trees #
-###################################
+#w/o diff
+all2: \
+	tree.default_upgma.nw
 
+data: ## Regenerate test data
+data: tree.precomputed.newick file_list.txt
+
+
+# Different ways to compute trees
 tree.default.nw: file_list.txt
+	@ >&2 printf "=====\\n$@\\n=====\\n"
 	../../attotree/attotree.py $(shell cat $<) > "$@.tmp"
 	mv "$@.tmp" "$@"
 
+tree.default_upgma.nw: file_list.txt
+	@ >&2 printf "=====\\n$@\\n=====\\n"
+	../../attotree/attotree.py $(shell cat $<) -m upgma> "$@.tmp"
+	mv "$@.tmp" "$@"
+
+tree.default_local_tmp_dir.nw: file_list.txt
+	@ >&2 printf "=====\\n$@\\n=====\\n"
+	../../attotree/attotree.py $(shell cat $<) -D -d tmp_dir > "$@.tmp"
+	mv "$@.tmp" "$@"
+
+tree.default_o-param.nw: file_list.txt
+	@ >&2 printf "=====\\n$@\\n=====\\n"
+	../../attotree/attotree.py $(shell cat $<) -o "$@.tmp"
+	mv "$@.tmp" "$@"
+
 tree.default_verbose_debug.nw: file_list.txt
+	@ >&2 printf "=====\\n$@\\n=====\\n"
 	../../attotree/attotree.py -V -D $(shell cat $<) > "$@.tmp"
 	mv "$@.tmp" "$@"
 
 tree.file_list.nw: file_list.txt
+	@ >&2 printf "=====\\n$@\\n=====\\n"
 	../../attotree/attotree.py -L "$<"  > "$@.tmp"
 	mv "$@.tmp" "$@"
 
 tree.file_list_proc_subst.nw: file_list.txt
+	@ >&2 printf "=====\\n$@\\n=====\\n"
 	../../attotree/attotree.py -L <(head -n 99999 "$<")  > "$@.tmp"
 	mv "$@.tmp" "$@"
 
+tree.file_list_proc_subst_two_lists.nw: file_list.txt
+	@ >&2 printf "=====\\n$@\\n=====\\n"
+	../../attotree/attotree.py -L <(head -n 2 "$<") <(tail -n+3 "$<")  > "$@.tmp"
+	mv "$@.tmp" "$@"
+
 tree.precomputed.newick: tree.default.nw
 	cp $< $@
 
-##############
-# Input data #
-##############
-
 file_list.txt: test_spneumo.tar.xz
 	tar xvf $<
 	tar -tf $< > file_list.txt.tmp
 	mv file_list.txt.tmp file_list.txt
 
+
+###############
+## Auxiliary ##
+###############
 help: ## Print help messages
 	@echo -e "$$(grep -hE '^\S*(:.*)?##' $(MAKEFILE_LIST) \
 		| sed \
@@ -67,6 +101,7 @@ help: ## Print help messages
 
 clean: ## Clean
 	rm -f *.nw *.tmp
+	rm -fr tmp_dir/*
 
 cleanall: clean ## Clean all
 	rm -f file_list.txt *.fa

diff --git a/tests/02_simple_tree/tmp_dir/.gitignore b/tests/02_simple_tree/tmp_dir/.gitignore
@@ -0,0 +1,3 @@
+*
+!.gitignore
+
diff --git a/tests/02_simple_tree/tree.precomputed.newick b/tests/02_simple_tree/tree.precomputed.newick
@@ -1 +1 @@
-((203692:0.00537,001334:0.00473):0.00060,(302649:0.00543,101058:0.00531):0.00062,403790:0.01403);
+((203692:0.00537,001334:0.00473):0.00060,(302649:0.00543,101058:0.00531):0.00062,403790:0.01403);