You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

282 lines
6.8KB

  1. import "./tasks/mapping.wdl" as mapping
  2. import "./tasks/Dedup.wdl" as Dedup
  3. import "./tasks/deduped_Metrics.wdl" as deduped_Metrics
  4. import "./tasks/Realigner.wdl" as Realigner
  5. import "./tasks/BQSR.wdl" as BQSR
  6. import "./tasks/Haplotyper_gVCF.wdl" as Haplotyper_gVCF
  7. import "./tasks/GVCFtyper.wdl" as GVCFtyper
  8. import "./tasks/split_gvcf_files.wdl" as split_gvcf_files
  9. import "./tasks/benchmark.wdl" as benchmark
  10. import "./tasks/multiqc.wdl" as multiqc
  11. import "./tasks/mendelian.wdl" as mendelian
  12. import "./tasks/merge_mendelian.wdl" as merge_mendelian
  13. import "./tasks/quartet_mendelian.wdl" as quartet_mendelian
  14. import "./tasks/fastqc.wdl" as fastqc
  15. import "./tasks/fastqscreen.wdl" as fastqscreen
  16. import "./tasks/extract_tables.wdl" as extract_tables
  17. import "./tasks/D5_D6.wdl" as D5_D6
  18. import "./tasks/merge_family.wdl" as merge_family
  19. workflow {{ project_name }} {
  20. File inputSamplesFile
  21. Array[Array[File]] inputSamples = read_tsv(inputSamplesFile)
  22. String SENTIEON_INSTALL_DIR
  23. String SENTIEONdocker
  24. String FASTQCdocker
  25. String FASTQSCREENdocker
  26. String QUALIMAPdocker
  27. String BENCHMARKdocker
  28. String MENDELIANdocker
  29. String DIYdocker
  30. String MULTIQCdocker
  31. String fasta
  32. File ref_dir
  33. File dbmills_dir
  34. String db_mills
  35. File dbsnp_dir
  36. String dbsnp
  37. File screen_ref_dir
  38. File fastq_screen_conf
  39. File benchmarking_dir
  40. String project
  41. String disk_size
  42. String BIGcluster_config
  43. String SMALLcluster_config
  44. scatter (quartet in inputSamples){
  45. call mapping.mapping as mapping {
  46. input:
  47. SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
  48. group=quartet[2],
  49. sample=quartet[2],
  50. pl="ILLUMINAL",
  51. fasta=fasta,
  52. ref_dir=ref_dir,
  53. fastq_1=quartet[0],
  54. fastq_2=quartet[1],
  55. docker=SENTIEONdocker,
  56. disk_size=disk_size,
  57. cluster_config=BIGcluster_config
  58. }
  59. call fastqc.fastqc as fastqc {
  60. input:
  61. read1=quartet[0],
  62. read2=quartet[1],
  63. docker=FASTQCdocker,
  64. cluster_config=BIGcluster_config,
  65. disk_size=disk_size
  66. }
  67. call fastqscreen.fastq_screen as fastqscreen {
  68. input:
  69. read1=quartet[0],
  70. read2=quartet[1],
  71. screen_ref_dir=screen_ref_dir,
  72. fastq_screen_conf=fastq_screen_conf,
  73. docker=FASTQSCREENdocker,
  74. cluster_config=BIGcluster_config,
  75. disk_size=disk_size
  76. }
  77. call Dedup.Dedup as Dedup {
  78. input:
  79. SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
  80. sorted_bam=mapping.sorted_bam,
  81. sorted_bam_index=mapping.sorted_bam_index,
  82. sample=quartet[2],
  83. docker=SENTIEONdocker,
  84. disk_size=disk_size,
  85. cluster_config=BIGcluster_config
  86. }
  87. call deduped_Metrics.deduped_Metrics as deduped_Metrics {
  88. input:
  89. SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
  90. fasta=fasta,
  91. ref_dir=ref_dir,
  92. Dedup_bam=Dedup.Dedup_bam,
  93. Dedup_bam_index=Dedup.Dedup_bam_index,
  94. sample=quartet[2],
  95. docker=SENTIEONdocker,
  96. disk_size=disk_size,
  97. cluster_config=BIGcluster_config
  98. }
  99. call Realigner.Realigner as Realigner {
  100. input:
  101. SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
  102. fasta=fasta,
  103. ref_dir=ref_dir,
  104. Dedup_bam=Dedup.Dedup_bam,
  105. Dedup_bam_index=Dedup.Dedup_bam_index,
  106. db_mills=db_mills,
  107. dbmills_dir=dbmills_dir,
  108. sample=quartet[2],
  109. docker=SENTIEONdocker,
  110. disk_size=disk_size,
  111. cluster_config=BIGcluster_config
  112. }
  113. call BQSR.BQSR as BQSR {
  114. input:
  115. SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
  116. fasta=fasta,
  117. ref_dir=ref_dir,
  118. realigned_bam=Realigner.realigner_bam,
  119. realigned_bam_index=Realigner.realigner_bam_index,
  120. db_mills=db_mills,
  121. dbmills_dir=dbmills_dir,
  122. dbsnp=dbsnp,
  123. dbsnp_dir=dbsnp_dir,
  124. sample=quartet[2],
  125. docker=SENTIEONdocker,
  126. disk_size=disk_size,
  127. cluster_config=BIGcluster_config
  128. }
  129. call Haplotyper_gVCF.Haplotyper_gVCF as Haplotyper_gVCF {
  130. input:
  131. SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
  132. fasta=fasta,
  133. ref_dir=ref_dir,
  134. recaled_bam=BQSR.recaled_bam,
  135. recaled_bam_index=BQSR.recaled_bam_index,
  136. sample=quartet[2],
  137. docker=SENTIEONdocker,
  138. disk_size=disk_size,
  139. cluster_config=BIGcluster_config
  140. }
  141. }
  142. call GVCFtyper.GVCFtyper as GVCFtyper {
  143. input:
  144. ref_dir=ref_dir,
  145. SENTIEON_INSTALL_DIR=SENTIEON_INSTALL_DIR,
  146. fasta=fasta,
  147. vcf=Haplotyper_gVCF.vcf,
  148. vcf_idx=Haplotyper_gVCF.vcf_idx,
  149. project=project,
  150. docker=SENTIEONdocker,
  151. cluster_config=BIGcluster_config,
  152. disk_size=disk_size
  153. }
  154. call split_gvcf_files.split_gvcf_files as split_gvcf_files {
  155. input:
  156. gvcf=GVCFtyper.gvcf,
  157. docker=DIYdocker,
  158. project=project,
  159. cluster_config=SMALLcluster_config,
  160. disk_size=disk_size
  161. }
  162. Array[File] single_gvcf = split_gvcf_files.splited_vcf
  163. scatter (idx in range(length(single_gvcf))) {
  164. call benchmark.benchmark as benchmark {
  165. input:
  166. vcf=single_gvcf[idx],
  167. benchmarking_dir=benchmarking_dir,
  168. ref_dir=ref_dir,
  169. fasta=fasta,
  170. docker=BENCHMARKdocker,
  171. cluster_config=BIGcluster_config,
  172. disk_size=disk_size,
  173. }
  174. }
  175. call multiqc.multiqc as multiqc {
  176. input:
  177. read1_zip=fastqc.read1_zip,
  178. read2_zip=fastqc.read2_zip,
  179. txt1=fastqscreen.txt1,
  180. txt2=fastqscreen.txt2,
  181. summary=benchmark.summary,
  182. docker=MULTIQCdocker,
  183. cluster_config=SMALLcluster_config,
  184. disk_size=disk_size
  185. }
  186. call extract_tables.extract_tables as extract_tables {
  187. input:
  188. fastqc=multiqc.fastqc,
  189. fastqscreen=multiqc.fastqscreen,
  190. hap=multiqc.hap,
  191. aln=deduped_Metrics.dedeuped_aln_metrics,
  192. quality_yield=deduped_Metrics.deduped_QualityYield,
  193. wgs_metrics=deduped_Metrics.deduped_wgsmetrics,
  194. docker=DIYdocker,
  195. cluster_config=SMALLcluster_config,
  196. disk_size=disk_size
  197. }
  198. Boolean sister_tag = read_boolean(split_gvcf_files.sister_tag)
  199. Boolean quartet_tag = read_boolean(split_gvcf_files.quartet_tag)
  200. if (sister_tag) {
  201. call D5_D6.D5_D6 as D5_D6 {
  202. input:
  203. splited_vcf=split_gvcf_files.splited_vcf,
  204. project=project,
  205. docker=DIYdocker,
  206. cluster_config=SMALLcluster_config,
  207. disk_size=disk_size,
  208. }
  209. }
  210. if (quartet_tag) {
  211. call merge_family.merge_family as merge_family {
  212. input:
  213. splited_vcf=split_gvcf_files.splited_vcf,
  214. project=project,
  215. docker=DIYdocker,
  216. cluster_config=SMALLcluster_config,
  217. disk_size=disk_size,
  218. }
  219. Array[File] family_vcfs = merge_family.family_vcf
  220. scatter (idx in range(length(family_vcfs))) {
  221. call mendelian.mendelian as mendelian {
  222. input:
  223. family_vcf=family_vcfs[idx],
  224. ref_dir=ref_dir,
  225. fasta=fasta,
  226. docker=MENDELIANdocker,
  227. cluster_config=BIGcluster_config,
  228. disk_size=disk_size
  229. }
  230. call merge_mendelian.merge_mendelian as merge_mendelian {
  231. input:
  232. D5_trio_vcf=mendelian.D5_trio_vcf,
  233. D6_trio_vcf=mendelian.D6_trio_vcf,
  234. family_vcf=family_vcfs[idx],
  235. docker=DIYdocker,
  236. cluster_config=SMALLcluster_config,
  237. disk_size=disk_size
  238. }
  239. }
  240. call quartet_mendelian.quartet_mendelian as quartet_mendelian {
  241. input:
  242. project_mendelian_summary=merge_mendelian.project_mendelian_summary,
  243. project=project,
  244. docker=DIYdocker,
  245. cluster_config=SMALLcluster_config,
  246. disk_size=disk_size
  247. }
  248. }
  249. }