Mercurial > repos > mheinzl > variant_analyzer2

comparison read2mut.py @ 77:1797e461d674 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
author mheinzl
date Mon, 29 Mar 2021 09:22:57 +0000
parents 56f271641828
children fdfe9a919ff7
comparison
equal deleted inserted replaced
76:56f271641828 77:1797e461d674
318 i = np.where(np.array(['#'.join(str(i) for i in z) 318 i = np.where(np.array(['#'.join(str(i) for i in z)
319 for z in zip(mut_array[:, 0], mut_array[:, 1], mut_array[:, 2], mut_array[:, 3])]) == key1)[0][0] 319 for z in zip(mut_array[:, 0], mut_array[:, 1], mut_array[:, 2], mut_array[:, 3])]) == key1)[0][0]
320 ref = mut_array[i, 2] 320 ref = mut_array[i, 2]
321 alt = mut_array[i, 3] 321 alt = mut_array[i, 3]
322 dcs_median = cvrg_dict[key1][2] 322 dcs_median = cvrg_dict[key1][2]
323 whole_array = pure_tags_dict_short[key1].keys() 323 whole_array = list(pure_tags_dict_short[key1].keys())
324 324
325 tier_dict[key1] = {} 325 tier_dict[key1] = {}
326 values_tier_dict = [("tier 1.1", 0), ("tier 1.2", 0), ("tier 2.1", 0), ("tier 2.2", 0), ("tier 2.3", 0), ("tier 2.4", 0), ("tier 2.5", 0), 326 values_tier_dict = [("tier 1.1", 0), ("tier 1.2", 0), ("tier 2.1", 0), ("tier 2.2", 0), ("tier 2.3", 0), ("tier 2.4", 0), ("tier 2.5", 0),
327 ("tier 3.1", 0), ("tier 3.2", 0), ("tier 4", 0), ("tier 5.1", 0), ("tier 5.2", 0), ("tier 5.3", 0), ("tier 5.4", 0), ("tier 5.5", 0), 327 ("tier 3.1", 0), ("tier 3.2", 0), ("tier 4", 0), ("tier 5.1", 0), ("tier 5.2", 0), ("tier 5.3", 0), ("tier 5.4", 0), ("tier 5.5", 0),
328 ("tier 6", 0), ("tier 7", 0)] 328 ("tier 6", 0), ("tier 7", 0)]
602 ratio_dist_start1 = ratio_dist_start2 = ratio_dist_start3 = ratio_dist_start4 = False 602 ratio_dist_start1 = ratio_dist_start2 = ratio_dist_start3 = ratio_dist_start4 = False
603 ratio_dist_end1 = ratio_dist_end2 = ratio_dist_end3 = ratio_dist_end4 = False 603 ratio_dist_end1 = ratio_dist_end2 = ratio_dist_end3 = ratio_dist_end4 = False
604 604
605 # mate 1 - SSCS ab 605 # mate 1 - SSCS ab
606 softclipped_idx1 = [True if re.search(r"^[0-9]+S", string) or re.search(r"S$", string) else False for string in cigars_dcs1] 606 softclipped_idx1 = [True if re.search(r"^[0-9]+S", string) or re.search(r"S$", string) else False for string in cigars_dcs1]
607 ratio1 = safe_div(sum(softclipped_idx1), float(len(softclipped_idx1))) >= threshold_reads 607 safe_div_result = safe_div(sum(softclipped_idx1), float(len(softclipped_idx1)))
608 if (safe_div_result == None):
609 ratio1 = False
610 else:
611 ratio1 = safe_div_result >= threshold_reads
608 if any(ij is True for ij in softclipped_idx1): 612 if any(ij is True for ij in softclipped_idx1):
609 softclipped_both_ends_idx1 = [True if (re.search(r"^[0-9]+S", string) and re.search(r"S$", string)) else False for string in cigars_dcs1] 613 softclipped_both_ends_idx1 = [True if (re.search(r"^[0-9]+S", string) and re.search(r"S$", string)) else False for string in cigars_dcs1]
610 softclipped_start1 = [int(string.split("S")[0]) if re.search(r"^[0-9]+S", string) else -1 for string in cigars_dcs1] 614 softclipped_start1 = [int(string.split("S")[0]) if re.search(r"^[0-9]+S", string) else -1 for string in cigars_dcs1]
611 softclipped_end1 = [int(re.split("[A-Z]", str(string))[-2]) if re.search(r"S$", string) else -1 for string in cigars_dcs1] 615 softclipped_end1 = [int(re.split("[A-Z]", str(string))[-2]) if re.search(r"S$", string) else -1 for string in cigars_dcs1]
612 dist_start_read1 = [(pos - soft) if soft != -1 else thr + 1000 for soft, pos in zip(softclipped_start1, pos_read1)] 616 dist_start_read1 = [(pos - soft) if soft != -1 else thr + 1000 for soft, pos in zip(softclipped_start1, pos_read1)]
622 ratio_dist_start1 = safe_div(sum([True if x <= thr else False for x in dist_start_read1]), float(sum(softclipped_idx1))) >= threshold_reads 626 ratio_dist_start1 = safe_div(sum([True if x <= thr else False for x in dist_start_read1]), float(sum(softclipped_idx1))) >= threshold_reads
623 ratio_dist_end1 = safe_div(sum([True if x <= thr else False for x in dist_end_read1]), float(sum(softclipped_idx1))) >= threshold_reads 627 ratio_dist_end1 = safe_div(sum([True if x <= thr else False for x in dist_end_read1]), float(sum(softclipped_idx1))) >= threshold_reads
624 628
625 # mate 1 - SSCS ba 629 # mate 1 - SSCS ba
626 softclipped_idx4 = [True if re.search(r"^[0-9]+S", string) or re.search(r"S$", string) else False for string in cigars_dcs4] 630 softclipped_idx4 = [True if re.search(r"^[0-9]+S", string) or re.search(r"S$", string) else False for string in cigars_dcs4]
627 ratio4 = safe_div(sum(softclipped_idx4), float(len(softclipped_idx4))) >= threshold_reads 631 safe_div_result = safe_div(sum(softclipped_idx4), float(len(softclipped_idx4)))
632 if (safe_div_result == None):
633 ratio4 = False
634 else:
635 ratio4 = safe_div_result >= threshold_reads
628 if any(ij is True for ij in softclipped_idx4): 636 if any(ij is True for ij in softclipped_idx4):
629 softclipped_both_ends_idx4 = [True if (re.search(r"^[0-9]+S", string) and re.search(r"S$", string)) else False for string in cigars_dcs4] 637 softclipped_both_ends_idx4 = [True if (re.search(r"^[0-9]+S", string) and re.search(r"S$", string)) else False for string in cigars_dcs4]
630 softclipped_start4 = [int(string.split("S")[0]) if re.search(r"^[0-9]+S", string) else -1 for string in cigars_dcs4] 638 softclipped_start4 = [int(string.split("S")[0]) if re.search(r"^[0-9]+S", string) else -1 for string in cigars_dcs4]
631 softclipped_end4 = [int(re.split("[A-Z]", str(string))[-2]) if re.search(r"S$", string) else -1 for string in cigars_dcs4] 639 softclipped_end4 = [int(re.split("[A-Z]", str(string))[-2]) if re.search(r"S$", string) else -1 for string in cigars_dcs4]
632 dist_start_read4 = [(pos - soft) if soft != -1 else thr + 1000 for soft, pos in zip(softclipped_start4, pos_read4)] 640 dist_start_read4 = [(pos - soft) if soft != -1 else thr + 1000 for soft, pos in zip(softclipped_start4, pos_read4)]
642 ratio_dist_start4 = safe_div(sum([True if x <= thr else False for x in dist_start_read4]), float(sum(softclipped_idx4))) >= threshold_reads 650 ratio_dist_start4 = safe_div(sum([True if x <= thr else False for x in dist_start_read4]), float(sum(softclipped_idx4))) >= threshold_reads
643 ratio_dist_end4 = safe_div(sum([True if x <= thr else False for x in dist_end_read4]), float(sum(softclipped_idx4))) >= threshold_reads 651 ratio_dist_end4 = safe_div(sum([True if x <= thr else False for x in dist_end_read4]), float(sum(softclipped_idx4))) >= threshold_reads
644 652
645 # mate 2 - SSCS ab 653 # mate 2 - SSCS ab
646 softclipped_idx2 = [True if re.search(r"^[0-9]+S", string) or re.search(r"S$", string) else False for string in cigars_dcs2] 654 softclipped_idx2 = [True if re.search(r"^[0-9]+S", string) or re.search(r"S$", string) else False for string in cigars_dcs2]
647 ratio2 = safe_div(sum(softclipped_idx2), float(len(softclipped_idx2))) >= threshold_reads 655 safe_div_result = safe_div(sum(softclipped_idx2), float(len(softclipped_idx2)))
656 if (safe_div_result == None):
657 ratio2 = False
658 else:
659 ratio2 = safe_div_result >= threshold_reads
648 if any(ij is True for ij in softclipped_idx2): 660 if any(ij is True for ij in softclipped_idx2):
649 softclipped_both_ends_idx2 = [True if (re.search(r"^[0-9]+S", string) and re.search(r"S$", string)) else False for string in cigars_dcs2] 661 softclipped_both_ends_idx2 = [True if (re.search(r"^[0-9]+S", string) and re.search(r"S$", string)) else False for string in cigars_dcs2]
650 softclipped_start2 = [int(string.split("S")[0]) if re.search(r"^[0-9]+S", string) else -1 for string in cigars_dcs2] 662 softclipped_start2 = [int(string.split("S")[0]) if re.search(r"^[0-9]+S", string) else -1 for string in cigars_dcs2]
651 softclipped_end2 = [int(re.split("[A-Z]", str(string))[-2]) if re.search(r"S$", string) else -1 for string in cigars_dcs2] 663 softclipped_end2 = [int(re.split("[A-Z]", str(string))[-2]) if re.search(r"S$", string) else -1 for string in cigars_dcs2]
652 dist_start_read2 = [(pos - soft) if soft != -1 else thr + 1000 for soft, pos in zip(softclipped_start2, pos_read2)] 664 dist_start_read2 = [(pos - soft) if soft != -1 else thr + 1000 for soft, pos in zip(softclipped_start2, pos_read2)]
662 ratio_dist_start2 = safe_div(sum([True if x <= thr else False for x in dist_start_read2]), float(sum(softclipped_idx2))) >= threshold_reads 674 ratio_dist_start2 = safe_div(sum([True if x <= thr else False for x in dist_start_read2]), float(sum(softclipped_idx2))) >= threshold_reads
663 ratio_dist_end2 = safe_div(sum([True if x <= thr else False for x in dist_end_read2]), float(sum(softclipped_idx2))) >= threshold_reads 675 ratio_dist_end2 = safe_div(sum([True if x <= thr else False for x in dist_end_read2]), float(sum(softclipped_idx2))) >= threshold_reads
664 676
665 # mate 2 - SSCS ba 677 # mate 2 - SSCS ba
666 softclipped_idx3 = [True if re.search(r"^[0-9]+S", string) or re.search(r"S$", string) else False for string in cigars_dcs3] 678 softclipped_idx3 = [True if re.search(r"^[0-9]+S", string) or re.search(r"S$", string) else False for string in cigars_dcs3]
667 ratio3 = safe_div(sum(softclipped_idx3), float(len(softclipped_idx3))) >= threshold_reads 679 safe_div_result = safe_div(sum(softclipped_idx3), float(len(softclipped_idx3)))
680 if (safe_div_result == None):
681 ratio3 = False
682 else:
683 ratio3 = safe_div_result >= threshold_reads
668 if any(ij is True for ij in softclipped_idx3): 684 if any(ij is True for ij in softclipped_idx3):
669 softclipped_both_ends_idx3 = [True if (re.search(r"^[0-9]+S", string) and re.search(r"S$", string)) else False for string in cigars_dcs3] 685 softclipped_both_ends_idx3 = [True if (re.search(r"^[0-9]+S", string) and re.search(r"S$", string)) else False for string in cigars_dcs3]
670 softclipped_start3 = [int(string.split("S")[0]) if re.search(r"^[0-9]+S", string) else -1 for string in cigars_dcs3] 686 softclipped_start3 = [int(string.split("S")[0]) if re.search(r"^[0-9]+S", string) else -1 for string in cigars_dcs3]
671 softclipped_end3 = [int(re.split("[A-Z]", str(string))[-2]) if re.search(r"S$", string) else -1 for string in cigars_dcs3] 687 softclipped_end3 = [int(re.split("[A-Z]", str(string))[-2]) if re.search(r"S$", string) else -1 for string in cigars_dcs3]
672 dist_start_read3 = [(pos - soft) if soft != -1 else thr + 1000 for soft, pos in zip(softclipped_start3, pos_read3)] 688 dist_start_read3 = [(pos - soft) if soft != -1 else thr + 1000 for soft, pos in zip(softclipped_start3, pos_read3)]