Mercurial > repos > artbio > small_rna_maps
changeset 18:2c95c899d0a4 draft
planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 06c6feefeaaa66f7dec1f6143a2c5aaf91836320
| author | artbio |
|---|---|
| date | Thu, 22 Nov 2018 03:07:41 -0500 |
| parents | b28dcd4051e8 |
| children | f33afecac67a |
| files | small_rna_maps.py small_rna_maps.xml test-data/bed1.bed test-data/bed2.bed test-data/bed3.bed |
| diffstat | 5 files changed, 325 insertions(+), 6 deletions(-) [+] |
line wrap: on
line diff
--- a/small_rna_maps.py Thu Nov 15 12:29:57 2018 -0500 +++ b/small_rna_maps.py Thu Nov 22 03:07:41 2018 -0500 @@ -19,6 +19,14 @@ the_parser.add_argument('--sample_names', dest='sample_names', required=True, nargs='+', help='list of sample names') + the_parser.add_argument('--bed', dest='bed', required=False, + help='Name of bed output must be specified\ + if --cluster option used') + the_parser.add_argument('--bed_skipcluster', dest='bed_skipcluster', + required=False, type=int, default=0, + help='Skip clusters of size equal or less than\ + specified integer in the bed output. \ + Default = 1') the_parser.add_argument('--outputs', nargs='+', action='store', help='list of two output paths (only two)') the_parser.add_argument('-M', '--plot_methods', nargs='+', action='store', @@ -269,27 +277,40 @@ line = [str(i) for i in line] out.write('\t'.join(line) + '\n') - def write_cluster_table(self, clustered_dic, out): + def write_cluster_table(self, clustered_dic, out, bedpath, skip): ''' Writer of a tabular file Dataset, Chromosome, Chrom_length, Coordinate, Polarity, <some mapped value> out is an *open* file handler + bed is an a file handler internal to the function ''' + bed = open(bedpath, 'w') for key in sorted(clustered_dic): start = clustered_dic[key][1][0] end = clustered_dic[key][1][1] size = end - start + 1 + if self.nostrand: + polarity = '.' + elif key[2] == 'F': + polarity = '+' + else: + polarity = '-' density = float(clustered_dic[key][0]) / size line = [self.sample_name, key[0], self.chromosomes[key[0]], key[1], key[2], clustered_dic[key][0], str(start) + "-" + str(end), str(size), str(density)] line = [str(i) for i in line] + if size > skip: + bedline = [key[0], str(start-1), str(end), 'cluster', '.', + polarity] + bed.write('\t'.join(bedline) + '\n') out.write('\t'.join(line) + '\n') + bed.close() def main(inputs, samples, methods, outputs, minsize, maxsize, cluster, - nostrand): + nostrand, bedfile=None, bed_skipcluster=0): for method, output in zip(methods, outputs): out = open(output, 'w') if method == 'Size': @@ -312,10 +333,10 @@ "Size": mapobj.compute_size, "cluster": mapobj.write_cluster_table} if cluster: - token["cluster"](mapobj.map_dict, out) + token["cluster"](mapobj.map_dict, out, bedfile, + bed_skipcluster) else: token[method](mapobj.map_dict, out) - # mapobj.compute_coverage(mapobj.map_dict, out) out.close() @@ -326,4 +347,5 @@ args.sample_names = [name + '_' + str(i) for i, name in enumerate(args.sample_names)] main(args.inputs, args.sample_names, args.plot_methods, args.outputs, - args.minsize, args.maxsize, args.cluster, args.nostrand) + args.minsize, args.maxsize, args.cluster, args.nostrand, args.bed, + args.bed_skipcluster)
--- a/small_rna_maps.xml Thu Nov 15 12:29:57 2018 -0500 +++ b/small_rna_maps.xml Thu Nov 22 03:07:41 2018 -0500 @@ -1,4 +1,4 @@ -<tool id="small_rna_maps" name="small_rna_maps" version="2.8.0"> +<tool id="small_rna_maps" name="small_rna_maps" version="2.9.0"> <description></description> <requirements> <requirement type="package" version="1.11.2=py27_0">numpy</requirement> @@ -38,11 +38,14 @@ --plot_methods 'Counts' --outputs '$output_tab' --cluster ${plots_options.cluster} + --bed '$output_bed' + --bed_skipcluster ${plots_options.skip_cluster} ${plots_options.strandness} && #else: --plot_methods '${plots_options.first_plot}' --outputs '$output_tab' && #end if + Rscript '$__tool_directory__'/small_rna_maps.r --first_dataframe '$output_tab' @@ -128,6 +131,8 @@ help="Sets the distance (in nt) below which reads are clustered to a single median position" /> <param name="strandness" argument="--nostrand" type="boolean" truevalue="--nostrand" falsevalue="" checked="false" label="Ignore polarity of reads ?" help="Set if you wish to cluster reads regardless of whether they are forward or reverse"/> + <param name="skip_cluster" type="integer" label="do not report cluster equal or less than the specified number of nucleotides" value="0" + help="Set the number of nucleotides in the cluster for which and below which a cluster is not reported in the bed output. Set to 0 (default) reports all including singlets" /> </when> </conditional> <conditional name="ylimits_cond"> @@ -144,6 +149,9 @@ <outputs> <data format="tabular" name="output_tab" label="$plots_options.first_plot dataframe" /> + <data format="bed" name="output_bed" label="bed file for clusters" > + <filter>plots_options['plots_options_selector'] == 'cluster'</filter> + </data> <data format="tabular" name="extra_output_tab" label="$plots_options.extra_plot dataframe"> <filter>plots_options['plots_options_selector'] == 'two_plot'</filter> </data> @@ -192,9 +200,11 @@ <param name="plots_options_selector" value="cluster" /> <param name="first_plot" value="Counts" /> <param name="cluster" value="5" /> + <param name="skip_cluster" value="0" /> <param name="strandness" value="false" /> <output file="clustering.tab" name="output_tab" /> <output file="clustering.pdf" name="output_pdf" /> + <output file="bed1.bed" name="output_bed" /> </test> <test> <!-- 3 --> <repeat name="series"> @@ -206,9 +216,27 @@ <param name="plots_options_selector" value="cluster" /> <param name="first_plot" value="Counts" /> <param name="cluster" value="5" /> + <param name="skip_cluster" value="0" /> <param name="strandness" value="true" /> <output file="clustering_unstranded.tab" name="output_tab" /> <output file="clustering_unstranded.pdf" name="output_pdf" /> + <output file="bed2.bed" name="output_bed" /> + </test> + <test> <!-- 3.1 --> + <repeat name="series"> + <param name="inputs" value="input1.bam" ftype="bam" /> + <param name="normalization" value="1.0" /> + </repeat> + <param name="minsize" value="0" /> + <param name="maxsize" value="10000" /> + <param name="plots_options_selector" value="cluster" /> + <param name="first_plot" value="Counts" /> + <param name="cluster" value="5" /> + <param name="skip_cluster" value="1" /> + <param name="strandness" value="false" /> + <output file="clustering.tab" name="output_tab" /> + <output file="clustering.pdf" name="output_pdf" /> + <output file="bed3.bed" name="output_bed" /> </test> <test> <!-- 4 --> <repeat name="series">
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/bed1.bed Thu Nov 22 03:07:41 2018 -0500 @@ -0,0 +1,118 @@ +FBtr0070001 0 1 cluster . + +FBtr0070001 6 18 cluster . + +FBtr0070001 26 34 cluster . + +FBtr0070001 71 72 cluster . + +FBtr0070533 0 1 cluster . + +FBtr0070533 11 38 cluster . + +FBtr0070533 71 72 cluster . + +FBtr0070603 0 42 cluster . + +FBtr0070603 50 56 cluster . + +FBtr0070603 71 72 cluster . + +FBtr0070604 0 1 cluster . + +FBtr0070604 17 22 cluster . + +FBtr0070604 29 32 cluster . + +FBtr0070604 56 57 cluster . + +FBtr0070604 71 72 cluster . + +FBtr0070911 0 1 cluster . + +FBtr0070911 14 15 cluster . + +FBtr0070911 37 38 cluster . + +FBtr0070911 72 73 cluster . + +FBtr0078490 0 1 cluster . + +FBtr0078490 12 18 cluster . + +FBtr0078490 24 28 cluster . + +FBtr0078490 37 38 cluster . + +FBtr0078490 71 72 cluster . + +FBtr0078580 0 31 cluster . + +FBtr0078580 51 52 cluster . + +FBtr0078580 71 72 cluster . + +FBtr0078790 0 1 cluster . + +FBtr0078790 16 17 cluster . + +FBtr0078790 32 33 cluster . + +FBtr0078790 41 52 cluster . + +FBtr0078790 68 69 cluster . - +FBtr0078790 72 73 cluster . + +FBtr0079064 0 3 cluster . + +FBtr0079064 32 33 cluster . + +FBtr0079064 51 52 cluster . + +FBtr0079064 71 72 cluster . + +FBtr0079090 0 1 cluster . + +FBtr0079090 25 26 cluster . + +FBtr0079090 32 33 cluster . + +FBtr0079090 52 53 cluster . + +FBtr0079090 55 56 cluster . - +FBtr0079090 71 72 cluster . + +FBtr0079338 0 1 cluster . + +FBtr0079338 11 17 cluster . + +FBtr0079338 24 25 cluster . + +FBtr0079338 41 46 cluster . + +FBtr0079338 72 73 cluster . + +FBtr0079528 0 18 cluster . + +FBtr0079528 27 28 cluster . + +FBtr0079528 34 37 cluster . + +FBtr0079528 50 51 cluster . + +FBtr0079528 70 71 cluster . + +FBtr0079596 0 19 cluster . + +FBtr0079596 52 54 cluster . + +FBtr0079596 72 73 cluster . + +FBtr0079677 0 5 cluster . + +FBtr0079677 51 53 cluster . + +FBtr0079677 71 72 cluster . + +FBtr0079690 0 1 cluster . + +FBtr0079690 21 27 cluster . + +FBtr0079690 32 33 cluster . + +FBtr0079690 71 72 cluster . + +FBtr0079692 0 1 cluster . + +FBtr0079692 17 18 cluster . + +FBtr0079692 24 25 cluster . + +FBtr0079692 31 32 cluster . + +FBtr0079692 72 73 cluster . + +FBtr0079693 0 1 cluster . + +FBtr0079693 24 25 cluster . + +FBtr0079693 71 72 cluster . + +FBtr0079694 0 1 cluster . + +FBtr0079694 17 18 cluster . + +FBtr0079694 51 52 cluster . + +FBtr0079694 71 72 cluster . + +FBtr0079702 0 1 cluster . + +FBtr0079702 18 19 cluster . + +FBtr0079702 55 56 cluster . + +FBtr0079702 71 72 cluster . + +FBtr0079728 0 1 cluster . + +FBtr0079728 7 8 cluster . + +FBtr0079728 18 19 cluster . + +FBtr0079728 32 33 cluster . + +FBtr0079728 55 56 cluster . + +FBtr0079728 71 72 cluster . + +FBtr0079729 0 1 cluster . + +FBtr0079729 12 13 cluster . + +FBtr0079729 51 57 cluster . + +FBtr0079729 71 72 cluster . + +FBtr0079752 0 1 cluster . + +FBtr0079752 6 12 cluster . + +FBtr0079752 32 33 cluster . + +FBtr0079752 51 52 cluster . + +FBtr0079752 71 72 cluster . + +FBtr0079820 0 1 cluster . + +FBtr0079820 44 56 cluster . + +FBtr0079820 73 74 cluster . + +FBtr0080609 0 20 cluster . + +FBtr0080609 41 42 cluster . + +FBtr0080609 50 52 cluster . + +FBtr0080609 71 72 cluster . + +FBtr0080644 0 12 cluster . + +FBtr0080644 71 72 cluster . + +FBtr0080646 0 7 cluster . + +FBtr0080646 71 72 cluster . + +FBtr0080647 0 12 cluster . + +FBtr0080647 71 72 cluster . + +FBtr0080660 0 7 cluster . + +FBtr0080660 71 72 cluster . + +FBtr0080663 0 17 cluster . + +FBtr0080663 25 26 cluster . + +FBtr0080663 31 32 cluster . + +FBtr0080663 49 51 cluster . + +FBtr0080663 70 71 cluster . + +FBtr0080664 0 21 cluster . + +FBtr0080664 27 38 cluster . + +FBtr0080664 49 56 cluster . + +FBtr0080664 70 71 cluster . +
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/bed2.bed Thu Nov 22 03:07:41 2018 -0500 @@ -0,0 +1,116 @@ +FBtr0070001 0 1 cluster . . +FBtr0070001 6 18 cluster . . +FBtr0070001 26 34 cluster . . +FBtr0070001 71 72 cluster . . +FBtr0070533 0 1 cluster . . +FBtr0070533 11 38 cluster . . +FBtr0070533 71 72 cluster . . +FBtr0070603 0 42 cluster . . +FBtr0070603 50 56 cluster . . +FBtr0070603 71 72 cluster . . +FBtr0070604 0 1 cluster . . +FBtr0070604 17 22 cluster . . +FBtr0070604 29 32 cluster . . +FBtr0070604 56 57 cluster . . +FBtr0070604 71 72 cluster . . +FBtr0070911 0 1 cluster . . +FBtr0070911 14 15 cluster . . +FBtr0070911 37 38 cluster . . +FBtr0070911 72 73 cluster . . +FBtr0078490 0 1 cluster . . +FBtr0078490 12 18 cluster . . +FBtr0078490 24 28 cluster . . +FBtr0078490 37 38 cluster . . +FBtr0078490 71 72 cluster . . +FBtr0078580 0 31 cluster . . +FBtr0078580 51 52 cluster . . +FBtr0078580 71 72 cluster . . +FBtr0078790 0 1 cluster . . +FBtr0078790 16 17 cluster . . +FBtr0078790 32 33 cluster . . +FBtr0078790 41 52 cluster . . +FBtr0078790 68 73 cluster . . +FBtr0079064 0 3 cluster . . +FBtr0079064 32 33 cluster . . +FBtr0079064 51 52 cluster . . +FBtr0079064 71 72 cluster . . +FBtr0079090 0 1 cluster . . +FBtr0079090 25 26 cluster . . +FBtr0079090 32 33 cluster . . +FBtr0079090 52 56 cluster . . +FBtr0079090 71 72 cluster . . +FBtr0079338 0 1 cluster . . +FBtr0079338 11 17 cluster . . +FBtr0079338 24 25 cluster . . +FBtr0079338 41 46 cluster . . +FBtr0079338 72 73 cluster . . +FBtr0079528 0 18 cluster . . +FBtr0079528 27 28 cluster . . +FBtr0079528 34 37 cluster . . +FBtr0079528 50 51 cluster . . +FBtr0079528 70 71 cluster . . +FBtr0079596 0 19 cluster . . +FBtr0079596 52 54 cluster . . +FBtr0079596 72 73 cluster . . +FBtr0079677 0 5 cluster . . +FBtr0079677 51 53 cluster . . +FBtr0079677 71 72 cluster . . +FBtr0079690 0 1 cluster . . +FBtr0079690 21 27 cluster . . +FBtr0079690 32 33 cluster . . +FBtr0079690 71 72 cluster . . +FBtr0079692 0 1 cluster . . +FBtr0079692 17 18 cluster . . +FBtr0079692 24 25 cluster . . +FBtr0079692 31 32 cluster . . +FBtr0079692 72 73 cluster . . +FBtr0079693 0 1 cluster . . +FBtr0079693 24 25 cluster . . +FBtr0079693 71 72 cluster . . +FBtr0079694 0 1 cluster . . +FBtr0079694 17 18 cluster . . +FBtr0079694 51 52 cluster . . +FBtr0079694 71 72 cluster . . +FBtr0079702 0 1 cluster . . +FBtr0079702 18 19 cluster . . +FBtr0079702 55 56 cluster . . +FBtr0079702 71 72 cluster . . +FBtr0079728 0 1 cluster . . +FBtr0079728 7 8 cluster . . +FBtr0079728 18 19 cluster . . +FBtr0079728 32 33 cluster . . +FBtr0079728 55 56 cluster . . +FBtr0079728 71 72 cluster . . +FBtr0079729 0 1 cluster . . +FBtr0079729 12 13 cluster . . +FBtr0079729 51 57 cluster . . +FBtr0079729 71 72 cluster . . +FBtr0079752 0 1 cluster . . +FBtr0079752 6 12 cluster . . +FBtr0079752 32 33 cluster . . +FBtr0079752 51 52 cluster . . +FBtr0079752 71 72 cluster . . +FBtr0079820 0 1 cluster . . +FBtr0079820 44 56 cluster . . +FBtr0079820 73 74 cluster . . +FBtr0080609 0 20 cluster . . +FBtr0080609 41 42 cluster . . +FBtr0080609 50 52 cluster . . +FBtr0080609 71 72 cluster . . +FBtr0080644 0 12 cluster . . +FBtr0080644 71 72 cluster . . +FBtr0080646 0 7 cluster . . +FBtr0080646 71 72 cluster . . +FBtr0080647 0 12 cluster . . +FBtr0080647 71 72 cluster . . +FBtr0080660 0 7 cluster . . +FBtr0080660 71 72 cluster . . +FBtr0080663 0 17 cluster . . +FBtr0080663 25 26 cluster . . +FBtr0080663 31 32 cluster . . +FBtr0080663 49 51 cluster . . +FBtr0080663 70 71 cluster . . +FBtr0080664 0 21 cluster . . +FBtr0080664 27 38 cluster . . +FBtr0080664 49 56 cluster . . +FBtr0080664 70 71 cluster . .
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/bed3.bed Thu Nov 22 03:07:41 2018 -0500 @@ -0,0 +1,35 @@ +FBtr0070001 6 18 cluster . + +FBtr0070001 26 34 cluster . + +FBtr0070533 11 38 cluster . + +FBtr0070603 0 42 cluster . + +FBtr0070603 50 56 cluster . + +FBtr0070604 17 22 cluster . + +FBtr0070604 29 32 cluster . + +FBtr0078490 12 18 cluster . + +FBtr0078490 24 28 cluster . + +FBtr0078580 0 31 cluster . + +FBtr0078790 41 52 cluster . + +FBtr0079064 0 3 cluster . + +FBtr0079338 11 17 cluster . + +FBtr0079338 41 46 cluster . + +FBtr0079528 0 18 cluster . + +FBtr0079528 34 37 cluster . + +FBtr0079596 0 19 cluster . + +FBtr0079596 52 54 cluster . + +FBtr0079677 0 5 cluster . + +FBtr0079677 51 53 cluster . + +FBtr0079690 21 27 cluster . + +FBtr0079729 51 57 cluster . + +FBtr0079752 6 12 cluster . + +FBtr0079820 44 56 cluster . + +FBtr0080609 0 20 cluster . + +FBtr0080609 50 52 cluster . + +FBtr0080644 0 12 cluster . + +FBtr0080646 0 7 cluster . + +FBtr0080647 0 12 cluster . + +FBtr0080660 0 7 cluster . + +FBtr0080663 0 17 cluster . + +FBtr0080663 49 51 cluster . + +FBtr0080664 0 21 cluster . + +FBtr0080664 27 38 cluster . + +FBtr0080664 49 56 cluster . +