Mercurial > repos > greg > ideas
comparison ideas.xml @ 165:bb5544d1c85e draft
Uploaded
author | greg |
---|---|
date | Thu, 18 Jan 2018 13:47:35 -0500 |
parents | 68d909206a15 |
children | 5c5e2f7b34c8 |
comparison
equal
deleted
inserted
replaced
164:68d909206a15 | 165:bb5544d1c85e |
---|---|
15 #set output_training_dir = "output_training_dir" | 15 #set output_training_dir = "output_training_dir" |
16 #set tmp_dir = "tmp" | 16 #set tmp_dir = "tmp" |
17 #set prep_input_config = "prep_input_config.txt" | 17 #set prep_input_config = "prep_input_config.txt" |
18 #set prep_output_config = "prep_output_config.txt" | 18 #set prep_output_config = "prep_output_config.txt" |
19 #set specify_genomic_window = $specify_genomic_window_cond.specify_genomic_window | 19 #set specify_genomic_window = $specify_genomic_window_cond.specify_genomic_window |
20 #set windows_positions_by_chroms_config = "windows_positions_by_chroms_config.txt" | |
20 #set perform_training = $perform_training_cond.perform_training | 21 #set perform_training = $perform_training_cond.perform_training |
21 ############################################## | 22 ############################################## |
22 ## Create the config file and prepare the data | 23 ## Create the config file and prepare the data |
23 ############################################## | 24 ############################################## |
24 #if str($output_heatmaps) == "yes": | 25 #if str($output_heatmaps) == "yes": |
25 mkdir '$output_pdf_dir' && | 26 mkdir '$output_pdf_dir' && |
26 #end if | 27 #end if |
27 #if str($perform_training) == "yes": | 28 #if str($perform_training) == "yes": |
29 #set output_dir = $output_training_dir | |
28 mkdir '$output_training_dir' && | 30 mkdir '$output_training_dir' && |
29 #end if | 31 #else: |
30 mkdir '$output_txt_dir' && | 32 #set output_dir = $output_txt_dir |
33 mkdir '$output_txt_dir' && | |
34 #end if | |
31 cp '$gen_prep_input_config' $prep_input_config && | 35 cp '$gen_prep_input_config' $prep_input_config && |
32 sort $prep_input_config -o $prep_input_config && | 36 sort $prep_input_config -o $prep_input_config && |
33 prepMat | 37 prepMat |
34 $prep_input_config | 38 $prep_input_config |
35 #if str($specify_genomic_window) == "yes": | 39 #if str($specify_genomic_window) == "yes": |
53 -exclude '$blacklist_input' | 57 -exclude '$blacklist_input' |
54 #end if | 58 #end if |
55 $norm | 59 $norm |
56 &>prepmat_log.txt; | 60 &>prepmat_log.txt; |
57 if [[ $? -ne 0 ]]; then | 61 if [[ $? -ne 0 ]]; then |
58 cp prepmat_log.txt '$output_txt_dir'; | 62 cp prepmat_log.txt '$output_dir'; |
59 exit 1; | 63 exit 1; |
60 fi | 64 fi |
61 ############################################## | 65 ############################################## |
62 ## Coerce the prepMat config output to the | 66 ## Coerce the prepMat config output to the |
63 ## format expected by IDEAS. | 67 ## format expected by IDEAS. |
69 ############################################## | 73 ############################################## |
70 ## Using a genomic window bed file, so categorize | 74 ## Using a genomic window bed file, so categorize |
71 ## the window positions by chromosome to enable | 75 ## the window positions by chromosome to enable |
72 ## the IDEAS -inv option. | 76 ## the IDEAS -inv option. |
73 ############################################## | 77 ############################################## |
74 #import collections | 78 && cp '$gen_windows_positions_by_chroms_config' $windows_positions_by_chroms_config |
75 #set window_positions_by_chroms = collections.OrderedDict() | 79 #end if |
76 #for count, line in enumerate(open($specify_genomic_window_cond.bed_input.file_name, 'r')): | 80 && Rscript '$__tool_directory__/ideas.R' |
77 #set $line = $line.strip() | 81 #if str($specify_genomic_window) == "yes": |
78 #if not $line or $line.startswith('#'): | 82 --windows_bed '$specify_genomic_window_cond.bed_input' |
79 #continue | 83 --windows_config $windows_positions_by_chroms_config |
80 #end if | 84 #end if |
81 #set items = $line.split('\t') | 85 #if str($perform_training) == "yes": |
82 #if $items[0] in $window_positions_by_chroms: | 86 --training_iterations $perform_training_cond.training_iterations |
83 #set tup = $window_positions_by_chroms[$items[0]] | 87 --training_windows $perform_training_cond.training_windows |
84 #set $tup[1] += 1 | 88 #end if |
85 #set $window_positions_by_chroms[$items[0]] = $tup | 89 --prep_output_config '$prep_output_config' |
86 #else: | 90 --hp $hp |
87 #set $window_positions_by_chroms[$items[0]] = [$count, $count+1] | 91 #if str($log2) != "0.0": |
88 #end if | 92 --log2 $log2 |
89 #end for | 93 #end if |
90 #for chrom, tup in $window_positions_by_chroms.items(): | 94 #if str($max_states) != "0.0": |
91 && Rscript '$__tool_directory__/ideas.R' | 95 --max_states $max_states |
92 #if str($perform_training) == "yes": | 96 #end if |
93 --training_iterations $perform_training_cond.training_iterations | 97 #if str($initial_states) != "0": |
94 --training_windows $perform_training_cond.training_windows | 98 --initial_states $initial_states |
95 #end if | 99 #end if |
96 --prep_output_config '$prep_output_config' | 100 #if str($max_position_classes) != "0": |
97 --windows_bed '$specify_genomic_window_cond.bed_input' | 101 --max_position_classes $max_position_classes |
98 --hp $hp | 102 #end if |
99 --window_start $tup[0] | 103 #if str($max_cell_type_clusters) != "0": |
100 --window_end $tup[1] | 104 --max_cell_type_clusters $max_cell_type_clusters |
101 #if str($log2) != "0.0": | 105 #end if |
102 --log2 $log2 | 106 #if str($prior_concentration) != "0.0": |
103 #end if | 107 --prior_concentration $prior_concentration |
104 #if str($max_states) != "0.0": | 108 #end if |
105 --max_states $max_states | 109 --burnin_num $burnin_num |
106 #end if | 110 --mcmc_num $mcmc_num |
107 #if str($initial_states) != "0": | 111 #if str($minerr) != "0.0": |
108 --initial_states $initial_states | 112 --minerr $minerr |
109 #end if | 113 #end if |
110 #if str($max_position_classes) != "0": | 114 #if str($maxerr) != "0.0": |
111 --max_position_classes $max_position_classes | 115 --maxerr $maxerr |
112 #end if | 116 #end if |
113 #if str($max_cell_type_clusters) != "0": | 117 --rseed $rseed |
114 --max_cell_type_clusters $max_cell_type_clusters | 118 --thread \${GALAXY_SLOTS:-4} |
115 #end if | 119 --project_name '$project_name' |
116 #if str($prior_concentration) != "0.0": | 120 #if str($save_ideas_log) == "yes": |
117 --prior_concentration $prior_concentration | 121 --save_ideas_log $save_ideas_log |
118 #end if | 122 --output_log '$output_log' |
119 --burnin_num $burnin_num | 123 #end if |
120 --mcmc_num $mcmc_num | 124 #if str($perform_training) == "yes": |
121 #if str($minerr) != "0.0": | 125 && mv ./*.para0 '$output_dir' |
122 --minerr $minerr | 126 && mv ./*.profile0 '$output_dir' |
123 #end if | |
124 #if str($maxerr) != "0.0": | |
125 --maxerr $maxerr | |
126 #end if | |
127 --rseed $rseed | |
128 --thread \${GALAXY_SLOTS:-4} | |
129 --project_name '$project_name.$chrom' | |
130 #if str($save_ideas_log) == "yes": | |
131 --save_ideas_log $save_ideas_log | |
132 --output_log '$output_log' | |
133 #end if | |
134 #end for | |
135 #else: | 127 #else: |
136 ############################################## | 128 && mv ./*.para '$output_dir' |
137 ## Not using a genomic window bed file. | 129 && mv ./*.profile '$output_dir' |
138 ############################################## | 130 #end if |
139 && Rscript '$__tool_directory__/ideas.R' | 131 && mv ./*.cluster '$output_dir' |
132 && mv ./*.state '$output_dir' | |
133 #if str($output_heatmaps) == "yes": | |
134 && Rscript '$__tool_directory__/create_heatmaps.R' | |
135 --input_dir '$output_dir' | |
136 --output_dir '$output_pdf_dir' | |
137 --script_dir '$__tool_directory__' | |
140 #if str($perform_training) == "yes": | 138 #if str($perform_training) == "yes": |
141 --training_iterations $perform_training_cond.training_iterations | |
142 --training_windows $perform_training_cond.training_windows | |
143 #end if | |
144 --prep_output_config '$prep_output_config' | |
145 --windows_bed $tmp_dir/*.bed | |
146 --hp $hp | |
147 #if str($log2) != "0.0": | |
148 --log2 $log2 | |
149 #end if | |
150 #if str($max_states) != "0.0": | |
151 --max_states $max_states | |
152 #end if | |
153 #if str($initial_states) != "0": | |
154 --initial_states $initial_states | |
155 #end if | |
156 #if str($max_position_classes) != "0": | |
157 --max_position_classes $max_position_classes | |
158 #end if | |
159 #if str($max_cell_type_clusters) != "0": | |
160 --max_cell_type_clusters $max_cell_type_clusters | |
161 #end if | |
162 #if str($prior_concentration) != "0.0": | |
163 --prior_concentration $prior_concentration | |
164 #end if | |
165 --burnin_num $burnin_num | |
166 --mcmc_num $mcmc_num | |
167 #if str($minerr) != "0.0": | |
168 --minerr $minerr | |
169 #end if | |
170 #if str($maxerr) != "0.0": | |
171 --maxerr $maxerr | |
172 #end if | |
173 --rseed $rseed | |
174 --thread \${GALAXY_SLOTS:-4} | |
175 --project_name '$project_name' | |
176 #if str($save_ideas_log) == "yes": | |
177 --save_ideas_log $save_ideas_log | |
178 --output_log '$output_log' | |
179 #end if | |
180 #end if | |
181 #if str($perform_training) == "yes": | |
182 && mv ./*.para0 '$output_training_dir' | |
183 && mv ./*.profile0 '$output_training_dir' | |
184 #if str($output_heatmaps) == "yes": | |
185 && Rscript '$__tool_directory__/create_heatmaps.R' | |
186 --input_dir 'output_training_dir' | |
187 --output_dir '$output_pdf_dir' | |
188 --script_dir '$__tool_directory__' | |
189 --in_training_mode true | 139 --in_training_mode true |
190 #end if | |
191 #else: | |
192 && mv ./*.cluster '$output_txt_dir' | |
193 && mv ./*.para '$output_txt_dir' | |
194 && mv ./*.profile '$output_txt_dir' | |
195 && mv ./*.state '$output_txt_dir' | |
196 #if str($output_heatmaps) == "yes": | |
197 && Rscript '$__tool_directory__/create_heatmaps.R' | |
198 --input_dir '$output_txt_dir' | |
199 --output_dir '$output_pdf_dir' | |
200 --script_dir '$__tool_directory__' | |
201 #end if | 140 #end if |
202 #end if | 141 #end if |
203 ]]></command> | 142 ]]></command> |
204 <configfiles> | 143 <configfiles> |
205 <configfile name="gen_prep_input_config"><![CDATA[#if str($cell_type_epigenetic_factor_cond.cell_type_epigenetic_factor) == "extract": | 144 <configfile name="gen_prep_input_config"><![CDATA[#if str($cell_type_epigenetic_factor_cond.cell_type_epigenetic_factor) == "extract": |
223 #else: | 162 #else: |
224 #for $input_items in $cell_type_epigenetic_factor_cond.input_repeat: | 163 #for $input_items in $cell_type_epigenetic_factor_cond.input_repeat: |
225 ${input_items.cell_type_name} ${input_items.epigenetic_factor_name} ${input_items.input} | 164 ${input_items.cell_type_name} ${input_items.epigenetic_factor_name} ${input_items.input} |
226 #end for | 165 #end for |
227 #end if]]></configfile> | 166 #end if]]></configfile> |
167 <configfile name="gen_windows_positions_by_chroms_config"><![CDATA[#if str($specify_genomic_window_cond.specify_genomic_window) == "yes": | |
168 #import collections | |
169 #set window_positions_by_chroms_odict = $collections.OrderedDict() | |
170 #for count, line in enumerate(open($specify_genomic_window_cond.bed_input.file_name, 'r')): | |
171 #set $line = $line.strip() | |
172 #if not $line or $line.startswith('#'): | |
173 #continue | |
174 #end if | |
175 #set items = $line.split('\t') | |
176 #if $items[0] in $window_positions_by_chroms_odict: | |
177 #set tup = $window_positions_by_chroms_odict[$items[0]] | |
178 #set $tup[1] += 1 | |
179 #set $window_positions_by_chroms_odict[$items[0]] = $tup | |
180 #else: | |
181 #set $window_positions_by_chroms_odict[$items[0]] = [$count, $count+1] | |
182 #end if | |
183 #end for | |
184 #for $chrom, $tup in $window_positions_by_chroms_odict.items(): | |
185 ${chrom} ${tup[0]} ${tup[1]} | |
186 #end for | |
187 #end if]]></configfile> | |
228 </configfiles> | 188 </configfiles> |
229 <inputs> | 189 <inputs> |
230 <conditional name="perform_training_cond"> | 190 <conditional name="perform_training_cond"> |
231 <param name="perform_training" type="select" label="Perform training?"> | 191 <param name="perform_training" type="select" label="Perform training?"> |
232 <option value="yes" selected="true">Yes</option> | 192 <option value="yes" selected="true">Yes</option> |
233 <option value="no">No</option> | 193 <option value="no">No</option> |
234 </param> | 194 </param> |
235 <when value="yes"> | 195 <when value="yes"> |
236 <param name="training_iterations" type="integer" value="20" min="2" label="Number of training iterations"/> | 196 <param name="training_iterations" type="integer" value="20" min="3" label="Number of training iterations"/> |
237 <param name="training_windows" type="integer" value="10000" min="2" label="Number of randomly selected windows for training"/> | 197 <param name="training_windows" type="integer" value="10000" min="2" label="Number of randomly selected windows for training"/> |
238 </when> | 198 </when> |
239 <when value="no"/> | 199 <when value="no"/> |
240 </conditional> | 200 </conditional> |
241 <conditional name="cell_type_epigenetic_factor_cond"> | 201 <conditional name="cell_type_epigenetic_factor_cond"> |