renluyao
/
repeat_validation
Vērot 1
Pievienot zvaigznīti 0
Atdalīts 0
Kods Problēmas 0 Izmaiņu pieprasījumi 0 Laidieni 0 Vikivietne Aktivitāte
Nevar pievienot vairāk kā 25 tēmas Tēmai ir jāsākas ar burtu vai ciparu, tā var saturēt domu zīmes ('-') un var būt līdz 35 simboliem gara.
5 Revīzijas
1 Atzars
Koks: 0f45a146c1
Atzari Tagi
${ item.name }
Izveidot atzaru ${ searchTerm }
no '0f45a146c1'
${ noResults }
repeat_validation/tasks/region.wdl

104 rindas
4.6KB
Neapstrādāts Vainot Vēsture 

  1. task region {

  2. 	File sub_vcf

  3. 	File sub_vcf_idx

  4. 

  5. 	File satellite

  6. 	File simple_repeat

  7. 	File SINE

  8. 	File LINE

  9. 	File LTR

  10. 	File low_complexity

  11. 	File SD

  12. 	File ins_breakpoint

  13. 	File del_breakpoint

  14. 	File mhc

  15. 

  16. 	String sample_name

  17. 	String tag

  18. 	String docker

  19. 	String cluster_config

  20. 	String disk_size

  21. 	

  22. 	command <<<

  23. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.satellite.vcf.gz --include-bed=${satellite}

  24. 

  25. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.simple_repeat.vcf.gz --include-bed=${simple_repeat}

  26. 

  27. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.SINE.vcf.gz --include-bed=${SINE}

  28. 

  29. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.LINE.vcf.gz --include-bed=${LINE}

  30. 

  31. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.LTR.vcf.gz --include-bed=${LTR}

  32. 

  33. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.low_complexity.vcf.gz --include-bed=${low_complexity}

  34. 

  35. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.SD.vcf.gz --include-bed=${SD}

  36. 

  37. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.ins_breakpoint.vcf.gz --include-bed=${ins_breakpoint}

  38. 

  39. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.del_breakpoint.vcf.gz --include-bed=${del_breakpoint}

  40. 

  41. 		rtg vcffilter -i ${sub_vcf} -o ${sample_name}.${tag}.mhc.vcf.gz --include-bed=${mhc}

  42. 

  43. 

  44. 		rtg vcfstat ${sample_name}.${tag}.satellite.vcf.gz > ${sample_name}.${tag}.satellite.vcfstat.txt

  45. 		rtg vcfstat ${sample_name}.${tag}.simple_repeat.vcf.gz > ${sample_name}.${tag}.simple_repeat.vcfstat.txt

  46. 		rtg vcfstat ${sample_name}.${tag}.SINE.vcf.gz > ${sample_name}.${tag}.SINE.vcfstat.txt

  47. 		rtg vcfstat ${sample_name}.${tag}.LINE.vcf.gz > ${sample_name}.${tag}.LINE.vcfstat.txt

  48. 		rtg vcfstat ${sample_name}.${tag}.LTR.vcf.gz > ${sample_name}.${tag}.LTR.vcfstat.txt

  49. 		rtg vcfstat ${sample_name}.${tag}.low_complexity.vcf.gz > ${sample_name}.${tag}.low_complexity.vcfstat.txt

  50. 		rtg vcfstat ${sample_name}.${tag}.SD.vcf.gz > ${sample_name}.${tag}.SD.vcfstat.txt

  51. 		rtg vcfstat ${sample_name}.${tag}.ins_breakpoint.vcf.gz > ${sample_name}.${tag}.ins_breakpoint.vcfstat.txt

  52. 		rtg vcfstat ${sample_name}.${tag}.del_breakpoint.vcf.gz > ${sample_name}.${tag}.del_breakpoint.vcfstat.txt

  53. 		rtg vcfstat ${sample_name}.${tag}.mhc.vcf.gz > ${sample_name}.${tag}.mhc.vcfstat.txt

  54. 

  55. 		for i in *vcf.gz

  56. 		do

  57. 			cat $i | grep -v '#' >> ${sample_name}.${tag}.all.count.txt

  58. 			cat $i | grep -v '#' | grep 'SNV;MC' | wc -l >> ${sample_name}.${tag}.snv.mc.count.txt

  59. 			cat $i | grep -v '#' | grep 'SNV;MC' | awk '{ if (($10 == $11) && ($11 == $12) && ($12 == $13)) { print } }' | wc -l >> ${sample_name}.${tag}.snv.mc.same.genotype.count.txt

  60. 			cat $i | grep -v '#' | grep 'INDEL;MC' | wc -l >> ${sample_name}.${tag}.indel.mc.count.txt

  61. 			cat $i | grep -v '#' | grep 'INDEL;MC' | awk '{ if (($10 == $11) && ($11 == $12) && ($12 == $13)) { print } }' | wc -l >> ${sample_name}.${tag}.indel.mc.same.genotype.count.txt

  62. 		done

  63. 

  64. 		ls *vcf.gz > file.list.txt

  65. 

  66. 

  67. 

  68. 	>>>

  69. 

  70. 	runtime {

  71. 		docker:docker

  72. 		cluster: cluster_config

  73. 		systemDisk: "cloud_ssd 40"

  74. 		dataDisk: "cloud_ssd " + disk_size + " /cromwell_root/"

  75. 	}

  76. 	output {

  77. 		File satellite_vcf = "${sample_name}.${tag}.satellite.vcf.gz"

  78. 		File simple_repeat_vcf = "${sample_name}.${tag}.simple_repeat.vcf.gz"

  79. 		File SINE_vcf = "${sample_name}.${tag}.SINE.vcf.gz"

  80. 		File LINE_vcf = "${sample_name}.${tag}.LINE.vcf.gz"

  81. 		File LTR_vcf = "${sample_name}.${tag}.LTR.vcf.gz"

  82. 		File low_complexity_vcf = "${sample_name}.${tag}.low_complexity.vcf.gz"

  83. 		File SD_vcf = "${sample_name}.${tag}.SD.vcf.gz"

  84. 		File ins_breakpoint_vcf = "${sample_name}.${tag}.ins_breakpoint.vcf.gz"

  85. 		File del_breakpoint_vcf = "${sample_name}.${tag}.del_breakpoint.vcf.gz"

  86. 		File mhc_vcf = "${sample_name}.${tag}.mhc.vcf.gz"

  87. 		File satellite_stat = "${sample_name}.${tag}.satellite.vcfstat.txt"

  88. 		File simple_repeat_stat = "${sample_name}.${tag}.simple_repeat.vcfstat.txt"

  89. 		File SINE_stat = "${sample_name}.${tag}.SINE.vcfstat.txt"

  90. 		File LINE_stat = "${sample_name}.${tag}.LINE.vcfstat.txt"

  91. 		File LTR_stat = "${sample_name}.${tag}.LTR.vcfstat.txt"

  92. 		File low_complexity_stat = "${sample_name}.${tag}.low_complexity.vcfstat.txt"

  93. 		File SD_stat = "${sample_name}.${tag}.SD.vcfstat.txt"

  94. 		File ins_breakpoint_stat = "${sample_name}.${tag}.ins_breakpoint.vcfstat.txt"

  95. 		File del_breakpoint_stat = "${sample_name}.${tag}.del_breakpoint.vcfstat.txt"

  96. 		File mhc_stat = "${sample_name}.${tag}.mhc.vcfstat.txt"

  97. 		File snv = "${sample_name}.${tag}.snv.mc.count.txt"

  98. 		File snv_mc = "${sample_name}.${tag}.snv.mc.same.genotype.count.txt"

  99. 		File indel = "${sample_name}.${tag}.indel.mc.count.txt"

  100. 		File indel_mc = "${sample_name}.${tag}.indel.mc.same.genotype.count.txt"

  101. 		File file_list = "file.list.txt"

  102. 		File all = "${sample_name}.${tag}.all.count.txt"

  103. 	}

  104. }