biopipen 0.28.1__py3-none-any.whl → 0.29.0__py3-none-any.whl

biopipen/utils/reference.py

@@ -1,5 +1,10 @@
+"""Utilities for indexing reference files"""
+from __future__ import annotations
+
 import tempfile
+from os import PathLike
 from pathlib import Path
+from typing import Literal
 
 from ..core.config import config
 from biopipen.utils.misc import run_command
@@ -18,7 +23,12 @@ def gztype(gzfile):
     return "flat"
 
 
-def tabix_index(infile, preset, tmpdir=None, tabix=config.exe.tabix):
+def tabix_index(
+    infile: str | PathLike,
+    preset: Literal["gff", "bed", "sam", "vcf", "gaf"],
+    tmpdir: bool | str | PathLike | None = None,
+    tabix: str = config.exe.tabix,
+) -> str | PathLike:
     """Index input file using tabix
 
     1. Try to check if there is an index file in the same directory where infile
@@ -28,22 +38,47 @@ def tabix_index(infile, preset, tmpdir=None, tabix=config.exe.tabix):
        it in tmpdir
     4. Index the bgzipped file and return the bgzipped file
 
+    Args:
+        infile: The input file to be indexed
+        preset: The preset used to index the file
+        tmpdir: The directory to link the infile there and index it
+            If False, try to index the infile directly. The directory where
+            the infile is should be writable.
+        tabix: The path to tabix
+
     Returns:
         The infile itself or re-bgzipped infile. This file comes with the
         index file in the same directory
     """
-    if tmpdir is None:
-        tmpdir = Path(tempfile.mkdtemp(prefix="biopipen_tabix_index_"))
-    else:
-        tmpdir = Path(tmpdir)
 
     infile = Path(infile)
     gt = gztype(infile)
-
-    if gt == "bgzip" and infile.with_suffix(infile.suffix + ".tbi").is_file():
+    index_file = infile.with_suffix(infile.suffix + ".tbi")
+    # if index file exists, and it's newer than the infile, return infile
+    if (
+        gt == "bgzip"
+        and index_file.is_file()
+        and index_file.stat().st_mtime > infile.resolve().stat().st_mtime
+    ):
         # only bgzipped file is possible to have index file
         return infile
 
+    if tmpdir is False:
+        # index the infile directly
+        run_command([tabix, "-p", preset, infile], fg=True)
+        return infile
+
+    if tmpdir is None:
+        from hashlib import md5
+        # use a hash of infile to create the tempdir
+        tmpdir = Path(tempfile.gettempdir()).joinpath(
+            f"biopipen_tabix_index_{md5(str(infile).encode()).hexdigest()}"
+        )
+    else:
+        tmpdir = Path(tmpdir)
+
+    tmpdir.mkdir(exist_ok=True, parents=True)
+
     # /path/to/some.vcf -> some.vcf
     # /path/to/some.vcf.gz -> some.vcf
     basename = infile.stem if infile.name.endswith(".gz") else infile.name
@@ -53,15 +88,24 @@ def tabix_index(infile, preset, tmpdir=None, tabix=config.exe.tabix):
     if gt == "gzip":
         # re-bgzip
         run_command(
-            ["gunzip", "-c", infile], stdout=new_infile.with_suffix(""),
+            ["gunzip", "-f", "-c", infile], stdout=new_infile.with_suffix(""),
         )
-        run_command(["bgzip", new_infile.with_suffix("")], fg=True)
+        run_command(["bgzip", "-f", new_infile.with_suffix("")], fg=True)
     elif gt == "flat":
-        run_command(["bgzip", "-c", infile], stdout=new_infile)
+        run_command(["bgzip", "-f", "-c", infile], stdout=new_infile)
     else:
+        if new_infile.is_symlink():
+            new_infile.unlink()
         # directory of infile may not have write permission
         new_infile.symlink_to(infile)
 
+    new_index_file = new_infile.with_suffix(new_infile.suffix + ".tbi")
+    if (
+        new_index_file.is_file()
+        and new_index_file.stat().st_mtime > infile.resolve().stat().st_mtime
+    ):
+        return new_infile
+
     run_command([tabix, "-p", preset, new_infile], fg=True)
     return new_infile
 

{biopipen-0.28.1.dist-info → biopipen-0.29.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: biopipen
-Version: 0.28.1
+Version: 0.29.0
 Summary: Bioinformatics processes/pipelines that can be run from `pipen run`
 License: MIT
 Author: pwwang
@@ -16,7 +16,7 @@ Provides-Extra: runinfo
 Requires-Dist: datar[pandas] (>=0.15.6,<0.16.0)
 Requires-Dist: pipen-board[report] (>=0.15,<0.16)
 Requires-Dist: pipen-cli-run (>=0.13,<0.14)
-Requires-Dist: pipen-filters (>=0.12,<0.13)
+Requires-Dist: pipen-filters (>=0.13,<0.14)
 Requires-Dist: pipen-poplog (>=0.1.2,<0.2.0)
 Requires-Dist: pipen-runinfo (>=0.6,<0.7) ; extra == "runinfo"
 Requires-Dist: pipen-verbose (>=0.11,<0.12)

{biopipen-0.28.1.dist-info → biopipen-0.29.0.dist-info}/RECORD

@@ -1,33 +1,33 @@
-biopipen/__init__.py,sha256=ZRQKbgDaGz_yuLk-cUKuk6ZBKCSRKZC8nQd041NRNXk,23
+biopipen/__init__.py,sha256=RvbohpnoJRSLAwJOFoWTtu5c5Vpi5vWAfsKR12hwJVI,23
 biopipen/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 biopipen/core/config.py,sha256=edK5xnDhM8j27srDzsxubi934NMrglLoKrdcC8qsEPk,1069
-biopipen/core/config.toml,sha256=20RCI30Peee1EQdfb_UbV3Hf74XUPndJnYZlUThytsw,1781
+biopipen/core/config.toml,sha256=IL31RfhuF-6V46lvLs1F-Z4SPmUuTvWzk5PN37Xjrqc,1907
 biopipen/core/defaults.py,sha256=yPeehPLk_OYCf71IgRVCWuQRxLAMixDF81Ium0HtPKI,344
 biopipen/core/filters.py,sha256=-OIzS5F_yNZ0Nk5Ci16BepPWqLOGBmYVqX3W98RSK9Y,12488
 biopipen/core/proc.py,sha256=60lUP3PcUAaKbDETo9N5PEIoeOYrLgcSmuytmrhcx8g,912
 biopipen/core/testing.py,sha256=lZ_R5ZbYPO2NPuLHdbzg6HbD_f4j8paVVbyeUqwg6FE,3411
 biopipen/ns/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-biopipen/ns/bam.py,sha256=5AsYrB0mtr_mH6mCL6gjJ5rC4NywpjFkpFjUrBGp7Fk,9301
-biopipen/ns/bcftools.py,sha256=puCDfIL-1z6cz2y1Rlz-ESNIr8xJgeIjEQ440qicCvM,3467
-biopipen/ns/bed.py,sha256=UN38qUChDeE-ipuSBY8RVLwvJqM2wxSRmlhOiDo4JG0,5395
+biopipen/ns/bam.py,sha256=jLhIztPiKsGfMpDgbcEpsybFskKkquIx9OKyTOc_L7Q,9172
+biopipen/ns/bed.py,sha256=HsTCJge7XNfCZyCBJ4iifNKQ5we4VZSpRx8XL8--y5A,6689
 biopipen/ns/cellranger.py,sha256=yPBoNzVSY74J7uyVucaob5lqZKKru5-hYSM4f4Nr2OY,5553
-biopipen/ns/cellranger_pipeline.py,sha256=lyRUdJEeQaTWGUz5UbrSdHMH_cgzXAXLlERmanppEBs,4037
-biopipen/ns/cnv.py,sha256=vq6dZfEOyuVuqg3nP6FQtNmQ-JocpBJMX9IYlZ0OPD0,6803
+biopipen/ns/cellranger_pipeline.py,sha256=EWkPJTujamNSMQoRnKfhUiIj6TkMfRmCSUbPfd8Tv8E,4011
+biopipen/ns/cnv.py,sha256=ssQAHf2MB675avoXVRkYy0vGiqIpRpRExywjhbymmBI,7811
 biopipen/ns/cnvkit.py,sha256=5mA2Q8-YDs4g1HoxtpB_NWnyZYwEThNr3s3wlubLQrQ,31130
-biopipen/ns/cnvkit_pipeline.py,sha256=2fJLn70L2jJ81ZMNdnU84Sf3HoKA2CSnHuDzLGR8jmw,36854
+biopipen/ns/cnvkit_pipeline.py,sha256=DZZljwvwVEI9sRUYBdxkOIT5-CIPLH1tag_OIU86aes,37094
 biopipen/ns/delim.py,sha256=fejsh4KW1TG5oMZzAC238LvQhBz7brXkfl3BHfnLK5M,5612
-biopipen/ns/gene.py,sha256=Q5FzRByfnRITXRNRZR65ApG09FRyiihRC3TcIXxufzE,2228
+biopipen/ns/gene.py,sha256=rty-Bjcf87v2vyb9X4kRvfrQ6XQYXgN4f2ftFO0nWA8,3888
 biopipen/ns/gsea.py,sha256=EsNRAPYsagaV2KYgr4Jv0KCnZGqayM209v4yOGGTIOI,7423
-biopipen/ns/misc.py,sha256=fzn0pXvdghMkQhu-e3MMapPNMyO6IAJbtTzVU3GbFa0,3246
-biopipen/ns/plot.py,sha256=fzJAKKl4a_tsVkLREGCQTFVHP049m33LdWgeYRb6v7M,5483
+biopipen/ns/misc.py,sha256=qXcm0RdR6W-xpYGgQn3v7JBeYRWwVm5gtgSj2tdVxx4,2935
+biopipen/ns/plot.py,sha256=PWPJg1Tly465drYxKMbfrFL4waWNrDG0plgYZeEA7Po,11972
+biopipen/ns/regulation.py,sha256=3esve77YuB-vYNoxYZl15GAd3HELX8FauBcrA2X3-mM,11750
 biopipen/ns/rnaseq.py,sha256=bKAa6friFWof4yDTWZQahm1MS-lrdetO1GqDKdfxXYc,7708
-biopipen/ns/scrna.py,sha256=OCEuE-TcqR7ADEEsgA6qL9rDK5R75YWRmmmvaNIFtZU,105925
+biopipen/ns/scrna.py,sha256=XQxdHDZjSt0TWTu-6mshYOmPwPg9a86siaYngBku90s,106954
 biopipen/ns/scrna_metabolic_landscape.py,sha256=6AhaynGG3lNRi96N2tReVT46BJMuEwooSSd2irBoN80,28347
-biopipen/ns/snp.py,sha256=EQ2FS0trQ7YThPmBVTpS66lc2OSfgQ6lCh6WnyP-C2g,5499
-biopipen/ns/stats.py,sha256=yJ6C1CXF84T7DDs9mgufqUOr89Rl6kybE5ji8Vnx6cw,13693
+biopipen/ns/snp.py,sha256=-Jx5Hsv_7KV7TqLU0nHCaPkMEN0CFdi4tNVlyq0rUZ4,27259
+biopipen/ns/stats.py,sha256=57UF9keLEGxg49R9C2Omx0WIXkPSHvLNxGxO3Xzap2M,16568
 biopipen/ns/tcgamaf.py,sha256=AFbUJIxiMSvsVY3RcHgjRFuMnNh2DG3Mr5slLNEyz6o,1455
 biopipen/ns/tcr.py,sha256=0PCF8iPZ629z6P3RHoAWEpMWmuDslomTGcMopjqvXmE,88304
-biopipen/ns/vcf.py,sha256=cdkKroii0_nl_bSP2cnO09qESUAhHqu6btOiTSKS79Y,15314
+biopipen/ns/vcf.py,sha256=0aKH_YSLy_-JzV-_VZb0EoScv7JKGrDU7BaeWHjDuRo,22699
 biopipen/ns/web.py,sha256=3zucrDo-IVsSnIvlw-deoScuxqWa6OMTm8Vo-R4E44Q,2224
 biopipen/reports/bam/CNAClinic.svelte,sha256=D4IxQcgDCPQZMbXog-aZP5iJEQTK2N4i0C60e_iXyfs,213
 biopipen/reports/bam/CNVpytor.svelte,sha256=s03SlhbEPd8-_44Dy_cqE8FSErhUdqStLK39te5o7ZE,1364
@@ -58,6 +58,11 @@ biopipen/reports/scrna_metabolic_landscape/MetabolicFeatures.svelte,sha256=4M8V6
 biopipen/reports/scrna_metabolic_landscape/MetabolicFeaturesIntraSubset.svelte,sha256=7FohWYR2BEqo77W4FuGsuwEEvG7j_Mc2ohXG8TSgODg,907
 biopipen/reports/scrna_metabolic_landscape/MetabolicPathwayActivity.svelte,sha256=bbPWDEXTi-A377Vo2VaEyh_VoxGmFj-P3WX8g_y_-9k,3416
 biopipen/reports/scrna_metabolic_landscape/MetabolicPathwayHeterogeneity.svelte,sha256=TohFc1IsHFAio8WNu_1FoaJucxsW3OtmS7fIJOH5KHc,350
+biopipen/reports/snp/PlinkCallRate.svelte,sha256=QoHqzJ88sx8d8FfH4ToFcR690RX0jmi2SaeWVAnw8y0,730
+biopipen/reports/snp/PlinkFreq.svelte,sha256=gHN1SezJ5l732Ndtrl_xZABdWPoSvAWkRh1rKbcsD1c,535
+biopipen/reports/snp/PlinkHWE.svelte,sha256=nes-NTwatoxvLoRBCg6jPp0fSg5kZZkVngURpfmM_nc,502
+biopipen/reports/snp/PlinkHet.svelte,sha256=hlSGx3qVETvZg6cd6vf27EtGeRzwD86I01kMLyIHEC4,545
+biopipen/reports/snp/PlinkIBD.svelte,sha256=s7nepF52Q4hKhRnIZDTFXqG7SIvhGy0NtfnTpexG0mE,507
 biopipen/reports/tcr/CDR3AAPhyschem.svelte,sha256=_r1pi_yTu_KjC02aegZ52IEQY8Mq0p3nqW_o9wOFWf8,1530
 biopipen/reports/tcr/CloneResidency.svelte,sha256=pzjkdGAhB7STxlXkTY5wnTT3plhkAPKMdGa6jwKuwoc,470
 biopipen/reports/tcr/Immunarch.svelte,sha256=jHOGUCWREaDNUf1_s6h9Kx_sp-y_dDRqY1rWrx0o0LM,433
@@ -72,47 +77,55 @@ biopipen/reports/vcf/TruvariConsistency.svelte,sha256=BBvtxi1EPmGH7j5M5zMOcLEhKW
 biopipen/scripts/bam/BamMerge.py,sha256=jyaJI0-TulAxaCzvrYjC8PujH_ECNInevnPbh2_XQtU,3477
 biopipen/scripts/bam/BamSplitChroms.py,sha256=b7GS2I4X0dLOhlPg_r9-buoIHTWlq6zHI3Rox94LXR8,4893
 biopipen/scripts/bam/CNAClinic.R,sha256=EbQ1s1_y2uCERTP1aP0Fwu_dO4n8k1YSMBswwVs0uDQ,5040
-biopipen/scripts/bam/CNVpytor.py,sha256=k7Ue_jh5Rp-qi6VDjkqLvbA47rMAyI-NBx7lv_otd2I,12893
+biopipen/scripts/bam/CNVpytor.py,sha256=hOUli9BDMOoth0or-tjUYC1AP3yNOuxUS6G3Rhcg99s,18000
 biopipen/scripts/bam/ControlFREEC.py,sha256=oX6iWsos-CfiT_ViDBrKeMOOIVdCKWrB-_MqzLgEF9s,3267
-biopipen/scripts/bcftools/BcftoolsAnnotate.py,sha256=0OIiP9bfR81vU1fbVgRPh8j860AEUPJ4yWFQvuZNiuI,1339
-biopipen/scripts/bcftools/BcftoolsFilter.py,sha256=P7OV9fwMSfFgD7ZtLWrU1-T2Oph94ACqtBMGi-SYrOk,2278
-biopipen/scripts/bcftools/BcftoolsSort.py,sha256=0xXIUf1Po8QHe8XTC5SyU_6kmLopRgp8Z87NXwWl8Mc,639
 biopipen/scripts/bed/Bed2Vcf.py,sha256=u0mp_2Y4UtEA839zq9UENesH6Gyiwd4sZQW9wFnBVU0,5090
 biopipen/scripts/bed/BedConsensus.py,sha256=gfAxuIalvCEpS0tiOyAJGPYGgHN0L-hm0K37Iteh5yw,2386
 biopipen/scripts/bed/BedLiftOver.sh,sha256=Y4gBsz9w4zhE29UmWojO6F4PXMMMWC1uCzjrxa19eOs,256
-biopipen/scripts/bed/BedtoolsMerge.py,sha256=TjKO5MpUzDj931bQAWku2660MVSiZzdMHt_v2Xbt0IE,355
+biopipen/scripts/bed/BedtoolsIntersect.py,sha256=sFyXPL3kG59xa5eJwHumcQLw3lfabOXsq2-k8IgIqt4,1722
+biopipen/scripts/bed/BedtoolsMerge.py,sha256=7mt307V_wWa_ME0VfuMsVX0HgEwfDcZtY_bDvOPjFiQ,368
 biopipen/scripts/cellranger/CellRangerCount.py,sha256=Mi6PiufFPp-cb4WRgZhniXi2ja5VUusRDW9RdAqFyX0,2831
 biopipen/scripts/cellranger/CellRangerSummary.R,sha256=8EqP4j074RLi5Adw47xyqQwkDEcvSS6Yfk0UgKKYOuo,11045
 biopipen/scripts/cellranger/CellRangerVdj.py,sha256=ipGkPzkiVy5IpPrYKUK4zgEIvz-aMJOKbMEN-bMStFE,2200
-biopipen/scripts/cnv/AneuploidyScore.R,sha256=liAN8u8_lj8voJ01oBW9Dw09yi388KF5f_gwPOv0wdE,8437
-biopipen/scripts/cnv/AneuploidyScoreSummary.R,sha256=9Zni5zqYfzevs5XSAt3fqD9WZ_RWr_ByUnXReKLLWoY,12337
-biopipen/scripts/cnv/TMADScore.R,sha256=uCLHQR6sMt-4uVUAEJlJxYXlai9ZE5J7xBl0sl-EkjU,1065
-biopipen/scripts/cnv/TMADScoreSummary.R,sha256=xcuEzHpzPNWWdZlurp5PS4D7665ZvIJtc4739pFACPI,7003
-biopipen/scripts/cnvkit/CNVkitAccess.py,sha256=l94f9L_C-tcNCFGCzZ7YQEZTBdpteY5Fq5DPAYu3zAI,645
-biopipen/scripts/cnvkit/CNVkitAutobin.py,sha256=Fy2f4IjNMaFyikdjSXEkkuf-bSwLggfwWpe4SOvZvPk,1593
-biopipen/scripts/cnvkit/CNVkitBatch.py,sha256=QhXwTkdJcP6woyuJgctJWaNURxJcb_xwID05lEapCcc,3775
+biopipen/scripts/cnv/AneuploidyScore.R,sha256=OldqRUr4rn8AZZCLDaX5xCIR1e0R7XXb_RQ2hXDHHog,9142
+biopipen/scripts/cnv/AneuploidyScoreSummary.R,sha256=fhffzQ5tW13KSgGz9CNDmvgqxUR1NDVVrsW_5xl1968,12410
+biopipen/scripts/cnv/TMADScore.R,sha256=IH07REfBO3QB2QNWuwGY6LgAUAOE0lsfP6W0bS67zGs,1530
+biopipen/scripts/cnv/TMADScoreSummary.R,sha256=w58LD5xHnLal9jv53nBUGpQ8eTkAyEk1leCVN2T_ZaM,7109
+biopipen/scripts/cnvkit/CNVkitAccess.py,sha256=UerX66utY8UdlI1mlSVD7E9O0ByLnHzt6G3Urb8mCWE,689
+biopipen/scripts/cnvkit/CNVkitAutobin.py,sha256=H5YcQ8f7Vs01Wad_oEQ9_AKBo5bKW856hAxShD75oDA,1656
+biopipen/scripts/cnvkit/CNVkitBatch.py,sha256=uzb300XLPRzV1BOOug6dAmupZ5FtQamFoYslVmbdTCA,3794
 biopipen/scripts/cnvkit/CNVkitCall.py,sha256=NyOnMjwT4w-ID9YtNks6fp6BIG7OzkGHmYphcXzAy_M,3544
-biopipen/scripts/cnvkit/CNVkitCoverage.py,sha256=t-PjHi0WujpAyhV_KgYIk4j6J-taY4ikdXJwdY96KSc,779
+biopipen/scripts/cnvkit/CNVkitCoverage.py,sha256=hTXh77QmMWy12Ipphdabkai19D0ZfKd7XF5Soc-HSG8,823
 biopipen/scripts/cnvkit/CNVkitDiagram.py,sha256=tvHW6wAZ9YXKb6sbdvWEO_HPjLK4LOYxrcnPkV8iUoA,1889
 biopipen/scripts/cnvkit/CNVkitFix.py,sha256=nZ7_15ba4zgTRpyQfyGAdcx1kCWyNTyjfEkKYxQumJA,1087
-biopipen/scripts/cnvkit/CNVkitGuessBaits.py,sha256=4CN7nx5FqWdw9coCPV1-RD-EM7kylnKR2qTtp_WoM7U,2197
-biopipen/scripts/cnvkit/CNVkitHeatmap.py,sha256=yv5oi2DkCIj-EI4PEBaazn4zRg7fJZEfdvGF8n2JlFQ,2478
-biopipen/scripts/cnvkit/CNVkitReference.py,sha256=_D0xefJm5DAFvS_omw6LLNyH7kChvixIKTgWjWkLyFc,1358
+biopipen/scripts/cnvkit/CNVkitGuessBaits.py,sha256=lrNNa3-jH87WCmdVv8lRmS894xT_zNjFjWXmi1oSG4U,2216
+biopipen/scripts/cnvkit/CNVkitHeatmap.py,sha256=gtLSI2ipnB9amEM_hJtl73iM-0jGG_fvn1uPMIlFa-o,2485
+biopipen/scripts/cnvkit/CNVkitReference.py,sha256=dO5vzqlgfecYUNJ3I2TvBdL_zKEJyKl3mRcL14GLDGU,1402
 biopipen/scripts/cnvkit/CNVkitScatter.py,sha256=7DhTiXPHEHbdXn0VFcDOR-wTP6sks08N94KG7zjo_Is,2728
 biopipen/scripts/cnvkit/CNVkitSegment.py,sha256=q5iGAjY6-yIehPcJpi3hX6EuGre0YgWTPkG_d5LEV48,1629
 biopipen/scripts/cnvkit/guess_baits.py,sha256=7OCMtSMHIJWWZv9qEYVXnB0N4hU_JaGEesKdkr6tvJc,10586
 biopipen/scripts/delim/RowsBinder.R,sha256=yp960u7Ui_jFCL8WDvODa-0vhJvyLo64ll35PzXYUbI,1444
 biopipen/scripts/delim/SampleInfo.R,sha256=1EYlqoVpIEl9l2eBaCLETuI_Ma3HjihS9tRbGmOPiBk,6397
-biopipen/scripts/gene/GeneNameConversion.py,sha256=2RveardTsLv2K1XSj3G0ERYLiln9bcR74bjkRdKcChc,1880
+biopipen/scripts/gene/GeneNameConversion.R,sha256=uAnueB1poicYpQHjUnl4ZBFn3sAWzP9HzmBC7zjlehI,1776
+biopipen/scripts/gene/GenePromoters.R,sha256=0ukq3-wpN6uq_cyJlQvKDVvy0Dv8cI5Htd9t2xpkEyk,2021
 biopipen/scripts/gsea/Enrichr.R,sha256=tr4vInlVIeiGXumh22ARuTQmy0-Qq869RiX7d7ERqCg,661
 biopipen/scripts/gsea/FGSEA.R,sha256=RLqDgrqnYEacHfzEEuZ3d29lxNqWehigOnGuu248SRg,1483
 biopipen/scripts/gsea/GSEA.R,sha256=gkiigYQ9xNdqZWMus0h0htSdWWbLJt9uXPo8Bc4hOaI,1083
 biopipen/scripts/gsea/PreRank.R,sha256=onZK1FQa6yDO0Fz4juy56XQjpzyw3zBdZv7edY9acNI,1356
 biopipen/scripts/misc/Config2File.py,sha256=NUio0uOEuZtUBpuByDSItYu9Kwu5mosb4pdPq5-QAmE,440
+biopipen/scripts/misc/Shell.sh,sha256=Rg8_DOfDFeUjZ_1_EVm2lbzyMLqcEv9A3k6SsTcfZZ0,493
 biopipen/scripts/misc/Str2File.py,sha256=99oQNxChxChNJ9vmD77b48cu-r_P_heSpx7A5wi3qTE,212
 biopipen/scripts/plot/Heatmap.R,sha256=4v_oRME8ZiwczIlBIp-OP_YPWLAvBKzbHiwNBCZ0Xog,1982
+biopipen/scripts/plot/Manhattan.R,sha256=14qWI4l4AgaEys59dI6nigZacQL_QAV9Hm4TVy_rlNY,4529
+biopipen/scripts/plot/QQPlot.R,sha256=qIvMzy3y1JIwsZDr-IlUxO4AexyakqYaMBNohIUv0i4,1678
 biopipen/scripts/plot/ROC.R,sha256=Cr-mHQx6c748fQYkOWO2xIKWwiVAUxGuxn6lYEhNH78,2430
 biopipen/scripts/plot/VennDiagram.R,sha256=GVc-kyHqnXrbXZvy-evcxI1XGtlLSChBiVnMjPywNMA,731
+biopipen/scripts/regulation/MotifAffinityTest.R,sha256=DcetElFSlurIRuOw1YLPvoOPDCnfq-EpgV-e1aKX_nM,8969
+biopipen/scripts/regulation/MotifAffinityTest_AtSNP.R,sha256=SAyTm2-6g5qVJFRrLxEY0QJrLWTkwDi_J_9J7HhtTN8,4438
+biopipen/scripts/regulation/MotifAffinityTest_MotifBreakR.R,sha256=wCK4tLx1iWh_OwW7ZvLTCjTGWCIfVqw-lYC0-JqIPqg,3338
+biopipen/scripts/regulation/MotifScan.py,sha256=nxwfDV3sXWk62x0WFpzxEZuUjJSErZvYkB2KJP7uAnc,5336
+biopipen/scripts/regulation/atSNP.R,sha256=TXJARbE0rDIzSq6Spacz_HsM_DXdREJ95ZsBg26trgw,1288
+biopipen/scripts/regulation/motifBreakR.R,sha256=trzvvTvwc5gSO427wkqx93FecuQxLGa9kNqodKa8S0U,70236
 biopipen/scripts/rnaseq/Simulation-ESCO.R,sha256=68cEHDdJclX8P8Q7ey9yBOfK09M_kxlL6zgYXsEL2Rs,6378
 biopipen/scripts/rnaseq/Simulation-RUVcorr.R,sha256=6C6Ke5RLF0fC2V9WQPoFEdqoDabCnhslZBIyB6zhIxc,1155
 biopipen/scripts/rnaseq/Simulation.R,sha256=PNEDUwEEOMGwgNyx5VPDAbzSLS74gqAxKenv5Uw_8zo,850
@@ -130,7 +143,7 @@ biopipen/scripts/scrna/ExprImputation-alra.R,sha256=w3W1txJcdWg52-SETY2Z0lO7maDN
 biopipen/scripts/scrna/ExprImputation-rmagic.R,sha256=jYIfqZpnvjKJkvItLnemPVtUApHBYQi1_L8rHVbEe1M,735
 biopipen/scripts/scrna/ExprImputation-scimpute.R,sha256=mg40qCUW7-nP5oHPvARq7dmtoahM0GRFWXQpum0BXVk,1082
 biopipen/scripts/scrna/ExprImputation.R,sha256=GcdZJpkDpq88hRQjtLZY5-byp8V43stEFm5T-pQbU6A,319
-biopipen/scripts/scrna/MarkersFinder.R,sha256=M7fHTbHHErZ9JbLmjDkx-6yVIay0_h0MkvgFegnqL44,22918
+biopipen/scripts/scrna/MarkersFinder.R,sha256=-rQyu0uXJscQbRUH6XDPxIFawOwqDxch75yyG1ArMaU,22670
 biopipen/scripts/scrna/MetaMarkers.R,sha256=9ve1X0TrDzS_ZEW6HtU3n8R-uPx7q-hYMMNFVDSE8wQ,11272
 biopipen/scripts/scrna/ModuleScoreCalculator.R,sha256=JSHd-_-KiFqW8avCGxgU4T-C5BtDr2u0kwIvEu2lFIg,4188
 biopipen/scripts/scrna/RadarPlots.R,sha256=TGPUTUcHOHgd9rsNtLYT-N6WHiFNDBZsiIoqkyAJh0A,13020
@@ -143,14 +156,14 @@ biopipen/scripts/scrna/SeuratClusterStats-hists.R,sha256=YhuD-GePjJPSkR0iLRgV_hi
 biopipen/scripts/scrna/SeuratClusterStats-ngenes.R,sha256=GVKIXFNS_syCuSN8oxoBkjxxAeI5LdSxh-qLVkUsbDA,2146
 biopipen/scripts/scrna/SeuratClusterStats-stats.R,sha256=bBbvNCvV6dZLg9zvhh2nR48_53md5A5UEqrPXD00MZU,9263
 biopipen/scripts/scrna/SeuratClusterStats.R,sha256=ouWoj7Q644uG3MUlT23AES8f74g38-jPtPhINSvoUas,1267
-biopipen/scripts/scrna/SeuratClustering.R,sha256=kAvQq3RV86_KSv9NlUtUeQrPKkbhSsnv6Q4DoiTu8M0,6403
+biopipen/scripts/scrna/SeuratClustering.R,sha256=xumBiqNnXcOn50tRmcqev9EBRUvf-ck9rEpAWeDXToE,7530
 biopipen/scripts/scrna/SeuratFilter.R,sha256=BrYK0MLdaTtQvInMaQsmOt7oH_hlks0M1zykkJtg2lM,509
 biopipen/scripts/scrna/SeuratLoading.R,sha256=ekWKnHIqtQb3kHVQiVymAHXXqiUxs6KKefjZKjaykmk,900
-biopipen/scripts/scrna/SeuratMap2Ref.R,sha256=_G8pG7NRV2GOFDzKBLY1nkXR0DO1c-6NkX990_hC8mk,9127
+biopipen/scripts/scrna/SeuratMap2Ref.R,sha256=qnFkOCudqC-j5uAszBssWM9Xg_vO3H9_Hj0n5wdOmBc,9757
 biopipen/scripts/scrna/SeuratMetadataMutater.R,sha256=Pp4GsF3hZ6ZC2vroC3LSBmVa4B1p2L3hbh981yaAIeQ,1093
 biopipen/scripts/scrna/SeuratPreparing.R,sha256=t6GOcc9ZNwpRLeES7uBWja9RF6u6k5I_TXcdK4Ve7d0,18683
 biopipen/scripts/scrna/SeuratSplit.R,sha256=vdK11V39_Uo_NaOh76QWCtxObGaEr5Ynxqq0hTiSvsU,754
-biopipen/scripts/scrna/SeuratSubClustering.R,sha256=L1SwKhNNKvsQGrcj0ZjScW9BLuvdO2pg7U48Ospsot8,6096
+biopipen/scripts/scrna/SeuratSubClustering.R,sha256=ooEQhUneYvSiRcwlqC9Mc8WE0g5pLEtKtQWYrKN3cck,8001
 biopipen/scripts/scrna/SeuratSubset.R,sha256=yVA11NVE2FSSw-DhxQcJRapns0tNNHdyDYi5epO6SKM,1776
 biopipen/scripts/scrna/SeuratTo10X.R,sha256=1mh1R0Qlo1iHVrpMLUXyLDOA92QKJ4GzTMURTFRqsWg,901
 biopipen/scripts/scrna/Subset10X.R,sha256=T2nJBTwOe12AIKC2FZsMSv6xx3s-67CYZokpz5wshqY,2679
@@ -161,12 +174,22 @@ biopipen/scripts/scrna_metabolic_landscape/MetabolicFeatures.R,sha256=sOnHSH0Ld-
 biopipen/scripts/scrna_metabolic_landscape/MetabolicFeaturesIntraSubset.R,sha256=RPSxEHWk50Fyw5YPBVVGeWwd55Hi0zjbipLuM5O4tjs,5465
 biopipen/scripts/scrna_metabolic_landscape/MetabolicPathwayActivity.R,sha256=95DLX1Rz0tobOuDZ8V9YdGgO0KiNthhccoeeOK21tno,16216
 biopipen/scripts/scrna_metabolic_landscape/MetabolicPathwayHeterogeneity.R,sha256=rQ9iwGh9FNRZlJJzM4QItdyXmebfzLAq05ZAjb1kGUw,9831
-biopipen/scripts/snp/MatrixEQTL.R,sha256=zOR_mhn_sUXuxqgV82TPvDp-E1i5aJVA45QixyRP8no,5930
+biopipen/scripts/snp/MatrixEQTL.R,sha256=Wr1m4RLcxisJgy5_7Tp_SoOoDZIBfRjHuUVnYLBWtAo,7186
+biopipen/scripts/snp/Plink2GTMat.py,sha256=ypxuT6YEBuuSk9TML9SHhVtA38GlZvB-3tQakJd8YLQ,5052
+biopipen/scripts/snp/PlinkCallRate.R,sha256=lkx93TyZjnvOWuq7W5zvdIlYmTC4n4ys52VymA7PvPM,6160
+biopipen/scripts/snp/PlinkFilter.py,sha256=eqC18ne3rPUzIJ_KjqvwQ79MyeMxe0oh4S9PnhfhK1E,3200
+biopipen/scripts/snp/PlinkFreq.R,sha256=Y8R-jE4upWe6B4yf7G9BCjEyv6HXItHOaUQzVInXL-8,10605
+biopipen/scripts/snp/PlinkFromVcf.py,sha256=TZaJ117Qp-gbqPX7WK6r7Zw8XHVXbRO6wb0V1NQouxw,2284
+biopipen/scripts/snp/PlinkHWE.R,sha256=Ncs4ctTDCnfGxs5HbyNxsNuq3Cn_Dwo1iFKdbj6GJSc,2201
+biopipen/scripts/snp/PlinkHet.R,sha256=r35hzA3A5lrqJN1kWxd2bYucTZd1eNd_vDjqhprur34,3005
+biopipen/scripts/snp/PlinkIBD.R,sha256=DJjmcwWYso8pY6HTKXLEIWsTurctknQxOQScG_ikBck,5640
 biopipen/scripts/snp/PlinkSimulation.py,sha256=mSSoGGG6sbEPBcUGdHgbebUrg4DiHeyNyc7jLPjV5pY,4169
+biopipen/scripts/snp/PlinkUpdateName.py,sha256=sYyb0ek46wRQfclFfoJEhQyQ-zWsFd2xpFHKw3jbsq4,4055
 biopipen/scripts/stats/ChowTest.R,sha256=4p7NULmfOZSfeBSQ04els0h3cXOK5yeCJJ4-gEBPOGk,3617
 biopipen/scripts/stats/DiffCoexpr.R,sha256=5hQDV2_7bKdKUsOGMZUa0GS5rc7kFspxonNyFEPmtbc,4516
 biopipen/scripts/stats/LiquidAssoc.R,sha256=s-XJbFoOfH4eWSkxbbOSHZ1x16lY0Sdod_V1KvSkM8k,3727
-biopipen/scripts/stats/MetaPvalue.R,sha256=c26lYC4rxQ3D7vRvsXJ4_M-QIYTDTV8AEjXrag2_srU,3957
+biopipen/scripts/stats/MetaPvalue.R,sha256=fnqayZeHd61cP9PxZZAebg5lE7JQgFG5MElCct43S1M,4012
+biopipen/scripts/stats/MetaPvalue1.R,sha256=falhaOeoa8E7ZbddXfGsSdim5P7eQvA7RFGD7XSrBUk,1733
 biopipen/scripts/tcgamaf/Maf2Vcf.py,sha256=Cxh7fiSNCxWDTfIJqZDOOnaSrw-85S_fH2U-PWY03hc,704
 biopipen/scripts/tcgamaf/MafAddChr.py,sha256=V10HMisl12O3ZfXuRmFNdy5p-3mr43WCvy0GHxSpwfA,494
 biopipen/scripts/tcgamaf/maf2vcf.pl,sha256=hJKcH-NbgWK6fmK7f3qex7ozJJl-PqCNPXqpwfcHwJg,22707
@@ -193,7 +216,7 @@ biopipen/scripts/tcr/ImmunarchFilter.R,sha256=o25O36FwH_0w6F8DFQ0SfpcwDzlzaGefXq
 biopipen/scripts/tcr/ImmunarchLoading.R,sha256=l_l-gojiCKI_MWgIUe2zG5boVtNipBv4rACRJEcrnFE,5734
 biopipen/scripts/tcr/ImmunarchSplitIdents.R,sha256=FGCeGV0uSmFU91lKkldUAeV4A2m3hHw5X4GNi8ffGzI,1873
 biopipen/scripts/tcr/SampleDiversity.R,sha256=jQ1OU3b8vswD8tZhLt3fkcqJKrl2bhQX0giHM2rXz3Y,2643
-biopipen/scripts/tcr/TCRClusterStats.R,sha256=aaY9w1GZWQLVQR7wl8POGCzzW7q0ge6cWxrzmqoWNA8,12743
+biopipen/scripts/tcr/TCRClusterStats.R,sha256=40e_XhrpyKkDv6yUTnU0wUnQiFdiHpk30Hr0pri3uuM,12864
 biopipen/scripts/tcr/TCRClustering.R,sha256=eflUsYfq4aEaX9BVL0MiB7lNlot_L-8VaReK516go84,9236
 biopipen/scripts/tcr/TCRDock.py,sha256=jjzxMWp-hs0LDtA1mVbiWDvUieSO7X-F9yeKGy1LSTM,3026
 biopipen/scripts/tcr/TESSA.R,sha256=bfOixWLZy8yi0MzXncP67KjtCukwXEzsK5fCdMzB5VM,6822
@@ -210,6 +233,10 @@ biopipen/scripts/tcr/TESSA_source/utility.R,sha256=6qbkMV7yp4bgQe718QiASQUAgOE2e
 biopipen/scripts/tcr/VJUsage.R,sha256=LjHEbAHW3WriCYiM9-T6Esd4jc6pnoiSxBKTN_YA490,437
 biopipen/scripts/tcr/immunarch-patched.R,sha256=6OAaV6arDVSI409VHuh1oMCuxlMQp-EaM-kkF-sTlCI,7007
 biopipen/scripts/tcr/vdjtools-patch.sh,sha256=rL5qp2S18CrpqduKkeR1HVmmuWhCVJOMKJXvWiKdYIc,566
+biopipen/scripts/vcf/BcftoolsAnnotate.py,sha256=iS-T6IhumqePW5kdyi_Tb6rubyIiCMjSTIRkGL_ad-U,2582
+biopipen/scripts/vcf/BcftoolsFilter.py,sha256=AdQMXFTNLCS5eqYWMNIMbkK8qXJ5j8GYm7HdPopVk0c,2573
+biopipen/scripts/vcf/BcftoolsSort.py,sha256=tU0pTrEIB-7x6iOSfU-KpYY1rEidi6Q4179NntY3cGc,3778
+biopipen/scripts/vcf/BcftoolsView.py,sha256=Sj3KkYPpwQFo5kmZC5MRxItrSE5KVZi0jNYrRFck3Ow,2465
 biopipen/scripts/vcf/TruvariBench.sh,sha256=80yLQ73OzSgsJ4ltzgpcWxYvvX1hFnCG8YSBhhhRQ9Y,765
 biopipen/scripts/vcf/TruvariBenchSummary.R,sha256=YlcD9qLtK3aeW3XeLx8pXTe7yOkbaYJyLHb2TYijm8A,1474
 biopipen/scripts/vcf/TruvariConsistency.R,sha256=dqoLHCzCeyJ3JWyDX8Q8MzIFwkVK8Jg71Ml45d96Cxg,2293
@@ -218,29 +245,30 @@ biopipen/scripts/vcf/VcfAnno.py,sha256=FW626rAs_WSU7AHQMKjfkYoByUGh_gVyJM97neGfO
 biopipen/scripts/vcf/VcfDownSample.sh,sha256=yBlyoA8xyHPkVc1XAs1biNZmN7rDR-CMz8qVh2GGvuk,939
 biopipen/scripts/vcf/VcfFilter.py,sha256=0clQi9XuN0B33xAquQ3e7ce1PZlzIOmOFctNA-dpdEw,1881
 biopipen/scripts/vcf/VcfFix.py,sha256=r2bgwwb2hAMr42wjpFZSFcUDxGllvhkOaRY_r9sKLNE,1304
-biopipen/scripts/vcf/VcfFix_utils.py,sha256=Ev_rvgLY9GF-aNntWbjXbl6eVlubw5uyHUD-g1c6Rno,2294
+biopipen/scripts/vcf/VcfFix_utils.py,sha256=Ur1-nFFYYjg7HYiK_C6Is_MTZXFULKRTwTGqBD8jzyU,2299
 biopipen/scripts/vcf/VcfIndex.py,sha256=kMOdB8O4UWvvxeEnAa3iZoH7_ySKHCJ0LOpEds0GXf4,635
 biopipen/scripts/vcf/VcfIntersect.py,sha256=6u9tLlfmEhlSMkYLXurVOO2uU94JwKRegSHhCzliHX0,735
 biopipen/scripts/vcf/VcfLiftOver.sh,sha256=-mb-aY0kbr62Audm8urEh3iLef_FREVWgt8hs3ELNK8,565
 biopipen/scripts/vcf/VcfSplitSamples.py,sha256=GraKi7WluzDAvVVGljwd3Yif6MriniF8stZOmllZtgk,1528
+biopipen/scripts/vcf/bcftools_utils.py,sha256=p91rZp0RG6AtbpYA_KuKCfa6T5aNKMq_rK95QPL7mMI,1514
 biopipen/scripts/web/Download.py,sha256=WKC_t5ZEeJoKFyY9XwksHARcMbKmHMcxNEUDLMGJ0Cc,924
 biopipen/scripts/web/DownloadList.py,sha256=cZvdi3LVzlATiTvAXe0uuDDXGqB5jcR2zHrMLCEb2U8,1130
 biopipen/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 biopipen/utils/caching.R,sha256=qANQqH8p-VpvD8V4VSoqSfp0TFr4esujC7x3OFZsJMw,1687
 biopipen/utils/common_docstrs.py,sha256=Ow-g-yS5P7DEO37cP1X-xioRbYWygfQHxIuLIaDdrjs,6288
-biopipen/utils/gene.R,sha256=BzAwlLA8hO12vF-3t6IwEuTEeLa_jBll4zm_5qe3qoE,1243
-biopipen/utils/gene.py,sha256=qE_BqTayrJWxRdniffhcz6OhZcw9GUoOrj2EtFWH9Gw,2246
+biopipen/utils/gene.R,sha256=EITppeN7dwXy7LnFELYHeRCSHImYno0JncLyVWEAXOM,3757
+biopipen/utils/gene.py,sha256=ToRgAPYc8E72Hm6uYi84ImQaRMPZp5rN7O4CxK7vCag,4477
 biopipen/utils/gsea.R,sha256=BUr4pwfo7ZyinOyNa4O4dDxA1c50qWb3lpph374Yw_A,9239
 biopipen/utils/io.R,sha256=jIYdqdn0iRWfQYAZa5CjXi3fikqmYvPPLIXhobRe8sw,537
-biopipen/utils/misc.R,sha256=Y9J8gZDuPPYFxYNQtDEvKAk5j-K8j_-n0DunYvu_Hv8,10671
-biopipen/utils/misc.py,sha256=KJziAFY4Kl-0ZsO93vteY9gRLZg9BSYig-TDocHY36k,3601
+biopipen/utils/misc.R,sha256=qTninxi9XAM_3QJqwThflGA-j7p1aKpQ9Pjmix6Iy3U,12059
+biopipen/utils/misc.py,sha256=nkQHa3GMc7Ce0N2GKmucbRc8BMXgZ2yr_SPbq9RYA4s,3740
 biopipen/utils/mutate_helpers.R,sha256=Bqy6Oi4rrPEPJw0Jq32bVAwwBfZv7JJL9jFcK5x-cek,17649
 biopipen/utils/plot.R,sha256=pzl37PomNeUZPxohHZ2w93j3Fc4T0Qrc62FF-9MTKdw,4417
-biopipen/utils/reference.py,sha256=6bPSwQa-GiDfr7xLR9a5T64Ey40y24yn3QfQ5wDFZkU,4420
+biopipen/utils/reference.py,sha256=oi5evicLwHxF0KAIPNZohBeHJLJQNWFJH0cr2y5pgcg,5873
 biopipen/utils/rnaseq.R,sha256=Ro2B2dG-Z2oVaT5tkwp9RHBz4dp_RF-JcizlM5GYXFs,1298
 biopipen/utils/single_cell.R,sha256=pJjYP8bIZpNAtTQ32rOXhZxaM1Y-6D-xUcK3pql9tbk,4316
 biopipen/utils/vcf.py,sha256=ajXs0M_QghEctlvUlSRjWQIABVF02wPdYd-0LP4mIsU,9377
-biopipen-0.28.1.dist-info/METADATA,sha256=UPLio_HdqVCyeNe2VCcUTZXz1CGakrQFH7_Dbb9qnHQ,882
-biopipen-0.28.1.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-biopipen-0.28.1.dist-info/entry_points.txt,sha256=wu70aoBcv1UahVbB_5237MY-9M9_mzqmWjDD-oi3yz0,621
-biopipen-0.28.1.dist-info/RECORD,,
+biopipen-0.29.0.dist-info/METADATA,sha256=dQ6UcVOLcX89F69jLAtXIczJo77lhkrSDdE5QrBgIJg,882
+biopipen-0.29.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+biopipen-0.29.0.dist-info/entry_points.txt,sha256=99uvED4sthSUVArFUMdMNLbYkVIEDLPUYg92JB22lt8,625
+biopipen-0.29.0.dist-info/RECORD,,

{biopipen-0.28.1.dist-info → biopipen-0.29.0.dist-info}/entry_points.txt

@@ -1,6 +1,5 @@
 [pipen_cli_run]
 bam=biopipen.ns.bam
-bcftools=biopipen.ns.bcftools
 bed=biopipen.ns.bed
 cellranger=biopipen.ns.cellranger
 cellranger_pipeline=biopipen.ns.cellranger_pipeline
@@ -12,6 +11,7 @@ gene=biopipen.ns.gene
 gsea=biopipen.ns.gsea
 misc=biopipen.ns.misc
 plot=biopipen.ns.plot
+regulation=biopipen.ns.regulation
 rnaseq=biopipen.ns.rnaseq
 scrna=biopipen.ns.scrna
 scrna_metabolic_landscape=biopipen.ns.scrna_metabolic_landscape

biopipen/ns/bcftools.py

@@ -1,111 +0,0 @@
-"""handling VCF files using bcftools"""
-from ..core.proc import Proc
-from ..core.config import config
-
-
-class BcftoolsAnnotate(Proc):
-    """Add or remove annotations from VCF files
-
-    Input:
-        infile: The input VCF file
-        annfile: The annotation file
-
-    Output:
-        outfile: The annotated VCF file
-
-    Envs:
-        bcftools: Path to bcftools
-        tabix: Path to tabix, used to index infile and annfile
-        annfile: The annotation file. If `in.annfile` is provided,
-            this is ignored
-        ncores: Number of cores (`--nthread`) to use
-        cols: Overwrite `-c/--columns`
-        header: Headers to be added
-        args: Other arguments for `bcftools annotate`
-    """
-    input = "infile:file, annfile:file"
-    output = "outfile:file:{{in.infile | basename}}"
-    lang = config.lang.python
-    envs = {
-        "bcftools": config.exe.bcftools,
-        "tabix": config.exe.tabix,
-        "annfile": "",
-        "cols": [],
-        "header": [],
-        "args": {},
-        "ncores": config.misc.ncores,
-    }
-    script = "file://../scripts/bcftools/BcftoolsAnnotate.py"
-
-
-class BcftoolsFilter(Proc):
-    """Apply fixed threshold filters to VCF files
-
-    Input:
-        infile: The input VCF file
-
-    Output:
-        outfile: The filtered VCF file. If the `in.infile` is gzipped, this is
-            gzipped as well.
-
-    Envs:
-        bcftools: Path to bcftools
-        ncores: Number of cores (`--nthread`) to use
-        keep: Whether we should keep the filtered variants or not.
-        args: Other arguments for `bcftools annotate`
-        ncores: `nthread`
-        tmpdir: Path to save the intermediate files
-            Since the filters need to be applied one by one by bcftools
-        includes: and
-        excludes: include/exclude only sites for which EXPRESSION is true.
-            See: https://samtools.github.io/bcftools/bcftools.html#expressions
-            If provided, `envs.args.include/exclude` will be ignored.
-            If `str`/`list` used, The filter names will be `Filter%d`
-            A dict is used when keys are filter names and values are expressions
-    """
-    input = "infile:file"
-    output = "outfile:file:{{in.infile | basename}}"
-    lang = config.lang.python
-    envs = {
-        "bcftools": config.exe.bcftools,
-        "keep": True,
-        "ncores": config.misc.ncores,
-        "includes": None,
-        "excludes": None,
-        "tmpdir": config.path.tmpdir,
-        "args": {},
-    }
-    script = "file://../scripts/bcftools/BcftoolsFilter.py"
-
-
-class BcftoolsSort(Proc):
-    """Sort VCF files
-
-    Input:
-        infile: The input VCF file
-
-    Output:
-        outfile: The sorted VCF file.
-
-    Envs:
-        bcftools: Path to bcftools
-        gz: Whether to gzip the output file
-        index: Whether to index the output file (tbi) (`envs.gz` forced to True)
-        tmpdir: Path to save the intermediate files
-        args: Other arguments for `bcftools sort`. For example `max-mem`.
-            See also https://samtools.github.io/bcftools/bcftools.html#sort
-    """
-    input = "infile:file"
-    output = (
-        "outfile:file:{{in.infile | stem0}}.vcf"
-        "{% if envs.gz or envs.index %}.gz{% endif %}"
-    )
-    lang = config.lang.python
-    envs = {
-        "bcftools": config.exe.bcftools,
-        "gz": True,
-        "index": True,
-        "tmpdir": config.path.tmpdir,
-        "args": {},
-    }
-    script = "file://../scripts/bcftools/BcftoolsSort.py"

biopipen/scripts/bcftools/BcftoolsAnnotate.py

@@ -1,42 +0,0 @@
-from os import path
-
-from biopipen.utils.reference import tabix_index
-from biopipen.utils.misc import dict_to_cli_args, run_command
-
-infile = {{in.infile | repr}}  # pyright: ignore
-annfile = {{(in.annfile or envs.annfile) | repr}}  # pyright: ignore
-outfile = {{out.outfile | repr}}  # pyright: ignore
-joboutdir = {{job.outdir | repr}}  # pyright: ignore
-bcftools = {{envs.bcftools | repr}}  # pyright: ignore
-tabix = {{envs.tabix | repr}}  # pyright: ignore
-ncores = {{envs.ncores | repr}}  # pyright: ignore
-cols = {{envs.cols | repr}}  # pyright: ignore
-header = {{envs.header | repr}}  # pyright: ignore
-args = {{envs.args | repr}}  # pyright: ignore
-
-args[""] = bcftools
-args["_"] = tabix_index(infile, "vcf")
-args["o"] = outfile
-args["threads"] = ncores
-
-if annfile:
-    abname = path.basename(annfile)
-    ext = path.splitext(
-        abname[:-3] if abname.endswith('.gz') else abname
-    )[-1][1:]
-    args["a"] = tabix_index(annfile, ext, tabix)
-
-if cols and isinstance(cols, list):
-    args["c"] = ",".join(cols)
-
-if header:
-    if not isinstance(header, list):
-        header = [header]
-
-    headerfile = path.join(joboutdir, "header.txt")
-    with open(headerfile, "w") as fh:
-        for head in header:
-            fh.write(f"{head}\n")
-    args["h"] = headerfile
-
-run_command(dict_to_cli_args(args, dashify=True), fg=True)

biopipen/scripts/bcftools/BcftoolsFilter.py

@@ -1,79 +0,0 @@
-import shutil
-from pathlib import Path
-from hashlib import md5
-
-from biopipen.core.filters import dict_to_cli_args, run_command
-
-infile = {{in.infile | repr}}  # pyright: ignore
-outfile = {{out.outfile | repr}}  # pyright: ignore
-bcftools = {{envs.bcftools | repr}}  # pyright: ignore
-keep = {{envs.keep | repr}}  # pyright: ignore
-args = {{envs.args | repr}}  # pyright: ignore
-ncores = {{envs.ncores | repr}}  # pyright: ignore
-tmpdir = {{envs.tmpdir | repr}}  # pyright: ignore
-includes = {{envs.includes | repr}}  # pyright: ignore
-excludes = {{envs.excludes | repr}}  # pyright: ignore
-
-args[""] = bcftools
-args["_"] = infile
-args["o"] = outfile
-args["threads"] = ncores
-if "O" not in args and "output-type" not in args:
-    args["O"] = "z" if infile.endswith(".gz") else "v"
-if "m" not in args and "mode" not in args:
-    args["m"] = "+"
-
-tmpdir = (
-    Path(tmpdir) / f"biopipen-bcftoolsfilter-{md5(infile.encode()).hexdigest()}"
-)
-tmpdir.mkdir(parents=True, exist_ok=True)
-# a.vcf.gz -> a
-# a.vcf -> a
-stem = Path(infile).stem
-if stem.endswith(".vcf"):
-    stem = stem[:-4]
-# .vcf.gz
-# .gz
-ext = Path(infile).name[len(stem):]
-
-FILTER_INDEX = [1]
-
-def handle_filter(vcf, fname, filt, flag):
-    print("- Handling filter ", fname, ": ", filt, " ...")
-
-    arguments = args.copy()
-    arguments[flag] = filt
-    arguments["_"] = vcf
-    arguments["o"] = tmpdir / f"{stem}.{fname}{ext}"
-    if keep:
-        arguments["s"] = fname
-
-    run_command(dict_to_cli_args(arguments, dashify=True), fg=True)
-    return arguments["o"]
-
-
-def normalize_expr(expr, flag):
-    out = {}
-    if not expr:
-        return out
-    if isinstance(expr, list):
-        for ex in expr:
-            out[f"FILTER{FILTER_INDEX[0]}"] = (ex, flag)
-            FILTER_INDEX[0] += 1
-    elif isinstance(expr, dict):
-        for name, ex in expr.items():
-            out[name] = (ex, flag)
-    else: # str
-        out[f"FILTER{FILTER_INDEX[0]}"] = (expr, flag)
-        FILTER_INDEX[0] += 1
-    return out
-
-includes = normalize_expr(includes, "include")
-excludes = normalize_expr(excludes, "exclude")
-includes.update(excludes)
-
-# bcftools can be only done once at one filter
-for fname, (filt, flag) in includes.items():
-    infile = handle_filter(infile, fname, filt, flag)
-
-shutil.copy2(infile, outfile)

biopipen/scripts/bcftools/BcftoolsSort.py

@@ -1,19 +0,0 @@
-from biopipen.utils.misc import run_command, dict_to_cli_args
-
-infile = {{in.infile | quote}}  # pyright: ignore
-outfile = {{out.outfile | quote}}  # pyright: ignore
-bcftools = {{envs.bcftools | quote}}  # pyright: ignore
-gz = {{envs.gz | repr}}  # pyright: ignore
-args = {{envs.args | repr}}  # pyright: ignore
-tmpdir = {{envs.tmpdir | quote}}  # pyright: ignore
-index = {{envs.index | repr}}  # pyright: ignore
-
-args[""] = bcftools
-args["_"] = infile
-args["o"] = outfile
-args["O"] = "z" if gz or index else "v"
-
-run_command(dict_to_cli_args(args, dashify=True), fg=True)
-
-if index:
-    run_command([bcftools, "index", outfile], fg=True)