biopipen 0.31.7__py3-none-any.whl → 0.32.1__py3-none-any.whl

biopipen/scripts/tcr/GIANA/query.py

@@ -5,220 +5,222 @@ import subprocess as sp
 import pandas as pd
 from GIANA4 import *
 
-def CreateReference(rFile, outdir='./', Vgene=True, ST=3):
+
+def CreateReference(rFile, outdir="./", Vgene=True, ST=3):
     ## convert input reference file into a python workplace
-    h=open(rFile)
-    alines=h.readlines()
-    ww=alines[0].strip().split('\t')
-    if not ww[0].startswith('C'):
+    h = open(rFile)
+    alines = h.readlines()
+    ww = alines[0].strip().split("\t")
+    if not ww[0].startswith("C"):
         ## header line
-        hline=alines[0]
-        alines=alines[1:]        
-    elif 'CDR3' in ww[0]:
-        hline=alines[0]
-        alines=alines[1:]
+        hline = alines[0]
+        alines = alines[1:]
+    elif "CDR3" in ww[0]:
+        hline = alines[0]
+        alines = alines[1:]
     else:
-        hline='CDR3\t'+'\t'.join(['Info'+str(x) for x in range(len(ww)-1)])        
-    seqs=[]
-    vgs=[]
-    infoList=[]
-    count=0
+        hline = "CDR3\t" + "\t".join(["Info" + str(x) for x in range(len(ww) - 1)])
+    seqs = []
+    vgs = []
+    infoList = []
+    count = 0
     for ll in alines:
-        ww=ll.strip().split('\t')
-        cdr3=ww[0]
-        if '*' in cdr3:
+        ww = ll.strip().split("\t")
+        cdr3 = ww[0]
+        if "*" in cdr3:
             continue
-        if '_' in cdr3:
+        if "_" in cdr3:
             continue
         seqs.append(ww[0])
         if Vgene:
             vgs.append(ww[1])
-            infoList.append('\t'.join(ww[1:]))
+            infoList.append("\t".join(ww[1:]))
         else:
-            infoList.append('\t'.join(ww[1:]))
-        count+=1
-    LD,VD, ID,SD= BuildLengthDict(seqs, vGene=vgs,INFO=infoList,sIDs=[x for x in range(len(seqs))])
+            infoList.append("\t".join(ww[1:]))
+        count += 1
+    LD, VD, ID, SD = BuildLengthDict(
+        seqs, vGene=vgs, INFO=infoList, sIDs=[x for x in range(len(seqs))]
+    )
     LDu_r, VDu_r, IDu_r, SDu_r = CollapseUnique(LD, VD, ID, SD)
-    flagLD_r={}
-    dMD_r={}
+    flagLD_r = {}
+    dMD_r = {}
     for kk in LDu_r:
-        vss=SDu_r[kk]
-        vInfo=IDu_r[kk]
-        flagL=[len(x)-1 for x in vInfo]
-        flagLD_r[kk]=flagL
-        dM=np.array([EncodingCDR3(x[ST:-2], M6, n0) for x in vss])
-        dM=dM.astype("float32")        
-        dMD_r[kk]=dM
-##    ff0=re.sub('.txt','',rFile)
-##    outfile=outdir+ff0+'_giana_ref.shelve'
-##    giana_shelf = shelve.open(outfile, 'n')
-##    giana_shelf['flagLD']=flagLD_r
-##    giana_shelf['dMD']=dMD_r
-##    giana_shelf['LDu']=LDu_r
-##    giana_shelf['VDu']=VDu_r
-##    giana_shelf['IDu']=IDu_r
-##    giana_shelf['SDu']=SDu_r
-##    giana_shelf.close()
+        vss = SDu_r[kk]
+        vInfo = IDu_r[kk]
+        flagL = [len(x) - 1 for x in vInfo]
+        flagLD_r[kk] = flagL
+        dM = np.array([EncodingCDR3(x[ST:-2], M6, n0) for x in vss])
+        dM = dM.astype("float32")
+        dMD_r[kk] = dM
+    ##    ff0=re.sub('.txt','',rFile)
+    ##    outfile=outdir+ff0+'_giana_ref.shelve'
+    ##    giana_shelf = shelve.open(outfile, 'n')
+    ##    giana_shelf['flagLD']=flagLD_r
+    ##    giana_shelf['dMD']=dMD_r
+    ##    giana_shelf['LDu']=LDu_r
+    ##    giana_shelf['VDu']=VDu_r
+    ##    giana_shelf['IDu']=IDu_r
+    ##    giana_shelf['SDu']=SDu_r
+    ##    giana_shelf.close()
     return [LDu_r, VDu_r, IDu_r, SDu_r, dMD_r]
 
-def MakeQuery(qFile, rData=[],dbFile=None, Vgene=True, thr=7, ST=3, thr_s=3.3):
+
+def MakeQuery(qFile, rData=[], dbFile=None, Vgene=True, thr=7, ST=3, thr_s=3.3):
     if dbFile is not None:
         with shelve.open(dbFile) as db:
             for key in db:
-                globals()[key]=db[key]
+                globals()[key] = db[key]
     else:
-        if len(rData)==0:
-            raise("Need to provide either a reference file or a shelve")
-        LDu_r=rData[0]
-        VDu_r=rData[1]
-        IDu_r=rData[2]
-        SDu_r=rData[3]
-        dMD_r=rData[4]
-    h=open(qFile)
-    alines=h.readlines()
-    ww=alines[0].strip().split('\t')
-    if not ww[0].startswith('C'):
+        if len(rData) == 0:
+            raise ("Need to provide either a reference file or a shelve")
+        LDu_r = rData[0]
+        VDu_r = rData[1]
+        IDu_r = rData[2]
+        SDu_r = rData[3]
+        dMD_r = rData[4]
+    h = open(qFile)
+    alines = h.readlines()
+    ww = alines[0].strip().split("\t")
+    if not ww[0].startswith("C"):
         ## header line
-        hline=alines[0]
-        alines=alines[1:]        
-    elif 'CDR3' in ww[0]:
-        hline=alines[0]
-        alines=alines[1:]
+        hline = alines[0]
+        alines = alines[1:]
+    elif "CDR3" in ww[0]:
+        hline = alines[0]
+        alines = alines[1:]
     else:
-        hline='CDR3\t'+'\t'.join(['Info'+str(x) for x in range(len(ww)-1)])        
-    seqs=[]
-    vgs=[]
-    infoList=[]
-    count=0
+        hline = "CDR3\t" + "\t".join(["Info" + str(x) for x in range(len(ww) - 1)])
+    seqs = []
+    vgs = []
+    infoList = []
+    count = 0
     for ll in alines:
-        ww=ll.strip().split('\t')
-        cdr3=ww[0]
-        if '*' in cdr3:
+        ww = ll.strip().split("\t")
+        cdr3 = ww[0]
+        if "*" in cdr3:
             continue
-        if '_' in cdr3:
+        if "_" in cdr3:
             continue
         seqs.append(ww[0])
         if Vgene:
             vgs.append(ww[1])
-            infoList.append('\t'.join(ww[1:]))
+            infoList.append("\t".join(ww[1:]))
         else:
-            infoList.append('\t'.join(ww[1:]))
-        count+=1
-    LD,VD, ID,SD= BuildLengthDict(seqs, vGene=vgs,INFO=infoList,sIDs=[x for x in range(len(seqs))])
+            infoList.append("\t".join(ww[1:]))
+        count += 1
+    LD, VD, ID, SD = BuildLengthDict(
+        seqs, vGene=vgs, INFO=infoList, sIDs=[x for x in range(len(seqs))]
+    )
     LDu, VDu, IDu, SDu = CollapseUnique(LD, VD, ID, SD)
-    tmpFile='tmp_query.txt'
-    g=open(tmpFile,'w')
+    tmpFile = "tmp_query.txt"
+    g = open(tmpFile, "w")
     for kk in LDu:
-        vss=SDu[kk]
-        vInfo=IDu[kk]
-        vss_r=SDu_r[kk]
-        vInfo_r=IDu_r[kk]
-        flagL=[len(x)-1 for x in vInfo]
-        dM_r=dMD_r[kk]
-        dM=np.array([EncodingCDR3(x[ST:-2], M6, n0) for x in vss])
-        dM=dM.astype("float32")
-        nq=dM.shape[0]
-        nr=dM_r.shape[0]
-        vssc=vss+vss_r
-        vInfoc=vInfo+vInfo_r
-        dMc=np.concatenate((dM, dM_r))
-        index = faiss.IndexFlatL2(Ndim*6)
+        vss = SDu[kk]
+        vInfo = IDu[kk]
+        vss_r = SDu_r[kk]
+        vInfo_r = IDu_r[kk]
+        flagL = [len(x) - 1 for x in vInfo]
+        dM_r = dMD_r[kk]
+        dM = np.array([EncodingCDR3(x[ST:-2], M6, n0) for x in vss])
+        dM = dM.astype("float32")
+        nq = dM.shape[0]
+        nr = dM_r.shape[0]
+        vssc = vss + vss_r
+        vInfoc = vInfo + vInfo_r
+        dMc = np.concatenate((dM, dM_r))
+        index = faiss.IndexFlatL2(Ndim * 6)
         index.add(dMc)
         D, I = index.search(dM, 2)
-        vv=np.where((D[0:nq,1]<=thr))[0]
-        flagL=np.array(flagL)
-        vv0=np.where((D[0:nq,1]>thr) & (flagL>0))[0]
-        curList=[]
+        vv = np.where((D[0:nq, 1] <= thr))[0]
+        flagL = np.array(flagL)
+        vv0 = np.where((D[0:nq, 1] > thr) & (flagL > 0))[0]
+        curList = []
         for v in vv0:
             for ii in range(len(vInfoc[v])):
-                line=vssc[v]+'\t'+vInfoc[v][ii]+'\t'+'query\n'
-                _=g.write(line)
+                line = vssc[v] + "\t" + vInfoc[v][ii] + "\t" + "query\n"
+                _ = g.write(line)
         for v in vv:
-            tmpI=I[v,]
+            tmpI = I[v,]
             if v not in tmpI:
-                tmpI[0]=v
-            idx1=tmpI[0]
-            idx2=tmpI[1]
-            c1=vssc[idx1]
-            c2=vssc[idx2]
-            info1=vInfoc[idx1]
-            info2=vInfoc[idx2]
+                tmpI[0] = v
+            idx1 = tmpI[0]
+            idx2 = tmpI[1]
+            c1 = vssc[idx1]
+            c2 = vssc[idx2]
+            info1 = vInfoc[idx1]
+            info2 = vInfoc[idx2]
             for tmpInfo in info1:
-                tup1=(c1, tmpInfo)
+                tup1 = (c1, tmpInfo)
                 if tup1 not in curList:
-                    if idx1<nq:
-                        line1=c1+'\t'+tmpInfo+'\t'+'query\n'
+                    if idx1 < nq:
+                        line1 = c1 + "\t" + tmpInfo + "\t" + "query\n"
                     else:
-                        line1=c1+'\t'+tmpInfo+'\t'+'ref\n'
-                    _=g.write(line1)
+                        line1 = c1 + "\t" + tmpInfo + "\t" + "ref\n"
+                    _ = g.write(line1)
                     curList.append(tup1)
             for tmpInfo in info2:
-                tup2=(c2, tmpInfo)
+                tup2 = (c2, tmpInfo)
                 if tup2 not in curList:
-                    if idx2<nq:
-                        line2=c2+'\t'+tmpInfo+'\t'+'query\n'
+                    if idx2 < nq:
+                        line2 = c2 + "\t" + tmpInfo + "\t" + "query\n"
                     else:
-                        line2=c2+'\t'+tmpInfo+'\t'+'ref\n'
-                    _=g.write(line2)
+                        line2 = c2 + "\t" + tmpInfo + "\t" + "ref\n"
+                    _ = g.write(line2)
                     curList.append(tup2)
     g.close()
-    cmd='python3 GIANA4.1.py -f tmp_query.txt -S '+str(thr_s) ## updated to GIANA4.1
-    p=sp.run(cmd, shell=True)
-
-def MergeExist(refClusterFile, outFile='queryFinal.txt',queryClusterFile='tmp_query--RotationEncodingBL62.txt', direction='q'):
+    cmd = "python3 GIANA4.1.py -f tmp_query.txt -S " + str(
+        thr_s
+    )  ## updated to GIANA4.1
+    p = sp.run(cmd, shell=True)
+
+
+def MergeExist(
+    refClusterFile,
+    outFile="queryFinal.txt",
+    queryClusterFile="tmp_query--RotationEncodingBL62.txt",
+    direction="q",
+):
     ## This function compare the query file with ref cluster file and merge the two based on shared TCRs
     ## If direction is 'q', the overlapping clusters will be added to the query file
     ## If direction is 'r', the overlapping and non-overlapping clusters will be added to the reference file
-    refT=pd.read_table(refClusterFile, skiprows=2, delimiter='\t', header=None)
-    queryT=pd.read_table(queryClusterFile, skiprows=2, delimiter='\t', header=None)
-    nq=queryT.shape[1]
-    nr=refT.shape[1]
-    if nr != nq-1:
+    refT = pd.read_table(refClusterFile, skiprows=2, delimiter="\t", header=None)
+    queryT = pd.read_table(queryClusterFile, skiprows=2, delimiter="\t", header=None)
+    nq = queryT.shape[1]
+    nr = refT.shape[1]
+    if nr != nq - 1:
         print("ERROR: Make sure reference and the query samples have the same columns!")
         print("No query file is generated.")
         return
-    gn=np.unique(queryT[1])
-    queryTs=pd.DataFrame([], columns=queryT.columns)
+    gn = np.unique(queryT[1])
+    queryTs = pd.DataFrame([], columns=queryT.columns)
     for nn in gn:
-            tmp_ddq=queryT.loc[np.where(queryT[1]==nn)[0],:]
-            cls_lab=np.unique(tmp_ddq[nq-1])
-            if len(cls_lab)==1:
-                if cls_lab[0]=='ref':
-                    continue
-            queryTs=queryTs.append(tmp_ddq)
-    queryTs.index=range(queryTs.shape[0])
-    keyr=refT[0]+'_'+refT[2]
-    keyq=queryTs[0]+'_'+queryTs[2]
-    vvr=np.where(queryTs[nq-1]=='ref')[0]
-    vvr_in=np.where(keyr.isin(keyq[vvr]))[0]
-    gn_r=list(refT.loc[vvr_in,1].drop_duplicates())
-    ddo=pd.DataFrame([], columns=refT.columns)
+        tmp_ddq = queryT.loc[np.where(queryT[1] == nn)[0], :]
+        cls_lab = np.unique(tmp_ddq[nq - 1])
+        if len(cls_lab) == 1:
+            if cls_lab[0] == "ref":
+                continue
+        queryTs = queryTs.append(tmp_ddq)
+    queryTs.index = range(queryTs.shape[0])
+    keyr = refT[0] + "_" + refT[2]
+    keyq = queryTs[0] + "_" + queryTs[2]
+    vvr = np.where(queryTs[nq - 1] == "ref")[0]
+    vvr_in = np.where(keyr.isin(keyq[vvr]))[0]
+    gn_r = list(refT.loc[vvr_in, 1].drop_duplicates())
+    ddo = pd.DataFrame([], columns=refT.columns)
     for nn in gn_r:
-        tmp_dd=refT.loc[np.where(refT[1]==nn)[0],:]
-        tmpkey=tmp_dd[0]+'_'+tmp_dd[2]
-        vv=np.where(keyq.isin(tmpkey))[0][0]
-        gq=queryTs[1][vv]
-        tmp_dd[1]=gq
-        ddo=ddo.append(tmp_dd)
-    if direction=='q':
-        ddo[nq-1]='ref'
+        tmp_dd = refT.loc[np.where(refT[1] == nn)[0], :]
+        tmpkey = tmp_dd[0] + "_" + tmp_dd[2]
+        vv = np.where(keyq.isin(tmpkey))[0][0]
+        gq = queryTs[1][vv]
+        tmp_dd[1] = gq
+        ddo = ddo.append(tmp_dd)
+    if direction == "q":
+        ddo[nq - 1] = "ref"
         ## remove groups that contain only ref group
-        queryTs=queryTs.append(ddo)
-        queryTs=queryTs.drop_duplicates()
-        queryTs.to_csv(outFile, sep='\t',header=False,index=False)
-#    queryTs.index=range(queryTs.shape[0])
-    if direction=='r':
+        queryTs = queryTs.append(ddo)
+        queryTs = queryTs.drop_duplicates()
+        queryTs.to_csv(outFile, sep="\t", header=False, index=False)
+    #    queryTs.index=range(queryTs.shape[0])
+    if direction == "r":
         ## to be developed
         pass
-
-
-
-
-
-
-
-
-
-
-
-

biopipen/scripts/tcr/TCRClustering.R

@@ -60,10 +60,12 @@ prepare_clustcr = function(clustcr_dir) {
     }
     clustcr_source = '
 import sys
-import pandas as pd
-import clustcr
 import atexit
 
+import pandas as pd
+from scipy import sparse as scipy_sparse
+
+
 @atexit.register
 def clustcr_exit():
     import pandas as pd
@@ -78,13 +80,32 @@ def clustcr_exit():
     sys.stderr.write(f"- sklearn: {sklearn.__version__}\\n")
     sys.stderr.write(f"- matplotlib: {matplotlib.__version__}\\n")
 
+
+# Monkey-patch scipy.sparse.isspmatrix to adopt latest scipy v1.14
+# If not, an error is raised:
+#   numpy.linalg.LinAlgError: 0-dimensional array given.
+#   Array must be at least two-dimensional
+scipy_sparse.isspmatrix = lambda x: isinstance(
+    x,
+    (
+        scipy_sparse.spmatrix,
+        scipy_sparse.csr_array,
+        scipy_sparse.csr_matrix,
+        scipy_sparse.csc_array,
+        scipy_sparse.csc_matrix,
+    ),
+)
+
+
+import clustcr  # noqa: #402
+
 clustcr_dir, clustcr_infile = sys.argv[1:3]
 cdr3df = pd.read_csv(clustcr_infile, index_col=None)
 cdr3 = cdr3df.iloc[:, 0]
 
-clustering = clustcr.Clustering(%s)
+clustering = clustcr.Clustering()
 output = clustering.fit(cdr3)
-output.clusters_df.to_csv(clustcr_dir + "/clusters.txt", sep="\\t", index=False)
+output.clusters_df.to_csv(clustcr_dir + "/clusters.txt", sep="\t", index=False)
 '
     clustcr_file = file.path(clustcr_dir, "_clustcr.py")
     cat(sprintf(clustcr_source, clustering_args), file=clustcr_file)

biopipen/utils/single_cell.R

@@ -1,7 +1,7 @@
 suppressPackageStartupMessages(library(rlang))
 suppressPackageStartupMessages(library(dplyr))
 suppressPackageStartupMessages(library(tidyr))
-suppressPackageStartupMessages(library(immunarch))
+try(suppressPackageStartupMessages(library(immunarch)))
 
 #' Expand a Immunarch object into cell-level
 #'
@@ -114,3 +114,94 @@ immdata_from_expanded <- function(
     )
     out
 }
+
+#' Convert Seurat object to Anndata
+#'
+#' @param sobjfile Seurat object file
+#' @param outfile Output file
+#' @param assay Assay to be used
+#'
+#' @export
+seurat_to_anndata <- function(sobjfile, outfile, assay = NULL, log_info, tmpdir = NULL, log_indent = "") {
+    library(Seurat)
+    library(SeuratDisk)
+    library(hdf5r)
+    if (endsWith(sobjfile, ".rds") || endsWith(sobjfile, ".RDS")) {
+        library(digest)
+
+        dig <- digest::digest(sobjfile, algo = "md5")
+        dig <- substr(dig, 1, 8)
+        assay_name <- ifelse(is.null(assay), "", paste0("_", assay))
+        tmpdir <- tmpdir %||% dirname(outfile)
+        dir.create(tmpdir, showWarnings = FALSE)
+        h5seurat_file <- file.path(
+            tmpdir,
+            paste0(
+                tools::file_path_sans_ext(basename(outfile)),
+                assay_name, ".", dig, ".h5seurat"
+            )
+        )
+        if (file.exists(h5seurat_file) &&
+            (file.mtime(h5seurat_file) < file.mtime(sobjfile))) {
+            file.remove(h5seurat_file)
+        }
+        if (!file.exists(h5seurat_file)) {
+            log_info("{log_indent}Reading RDS file ...")
+            sobj <- readRDS(sobjfile)
+            assay <- assay %||% DefaultAssay(sobj)
+            # In order to convert to h5ad
+            # https://github.com/satijalab/seurat/issues/8220#issuecomment-1871874649
+            sobj$RNAv3 <- as(object = sobj[[assay]], Class = "Assay")
+            DefaultAssay(sobj) <- "RNAv3"
+            sobj$RNA <- NULL
+            sobj <- RenameAssays(sobj, RNAv3 = "RNA")
+
+            log_info("{log_indent}Saving to H5Seurat file ...")
+            SaveH5Seurat(sobj, h5seurat_file)
+            rm(sobj)
+            gc()
+            sobjfile <- h5seurat_file
+        } else {
+            log_info("{log_indent}Using existing H5Seurat file ...")
+        }
+    }
+
+    if (!endsWith(sobjfile, ".h5seurat")) {
+        stop(paste0("Unknown input file format: ",
+            tools::file_ext(sobjfile),
+            ". Supported formats: .rds, .RDS, .h5seurat"))
+    }
+
+    log_info("{log_indent}Converting to Anndata ...")
+    Convert(sobjfile, dest = outfile, assay = assay %||% "RNA", overwrite = TRUE)
+
+    log_info("{log_indent}Fixing categorical data ...")
+    # See: https://github.com/mojaveazure/seurat-disk/issues/183
+    H5.create_reference <- function(self, ...) {
+        space <- self$get_space()
+        do.call("[", c(list(space), list(...)))
+        ref_type <- hdf5r::h5const$H5R_OBJECT
+        ref_obj <- hdf5r::H5R_OBJECT$new(1, self)
+        res <- .Call("R_H5Rcreate", ref_obj$ref, self$id, ".", ref_type,
+                    space$id, FALSE, PACKAGE = "hdf5r")
+        if (res$return_val < 0) {
+            stop("Error creating object reference")
+        }
+        ref_obj$ref <- res$ref
+        return(ref_obj)
+    }
+
+    h5ad <- H5File$new(outfile, "r+")
+    cats <- names(h5ad[["obs/__categories"]])
+    for (cat in cats) {
+        catname <- paste0("obs/__categories/", cat)
+        obsname <- paste0("obs/", cat)
+        ref <- H5.create_reference(h5ad[[catname]])
+        h5ad[[obsname]]$create_attr(
+            attr_name = "categories",
+            robj = ref,
+            space = H5S$new(type = "scalar")
+        )
+    }
+    h5ad$close()
+}

{biopipen-0.31.7.dist-info → biopipen-0.32.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: biopipen
-Version: 0.31.7
+Version: 0.32.1
 Summary: Bioinformatics processes/pipelines that can be run from `pipen run`
 License: MIT
 Author: pwwang
@@ -12,6 +12,7 @@ Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
 Provides-Extra: runinfo
 Requires-Dist: datar[pandas] (>=0.15.6,<0.16.0)
 Requires-Dist: pipen-board[report] (>=0.16,<0.17)