20d541d28092f1324307a0d34a1e65d145be6446 max Thu Jun 26 04:46:08 2025 -0700 removing config filename checks, refs #32013 diff --git src/cbPyLib/cellbrowser/cellbrowser.py src/cbPyLib/cellbrowser/cellbrowser.py index 512a4f5..cfec862 100755 --- src/cbPyLib/cellbrowser/cellbrowser.py +++ src/cbPyLib/cellbrowser/cellbrowser.py @@ -1569,31 +1569,30 @@ def iterRows(self): " yield (geneId, symbol, array) tuples from gene expression file. " mat = self.mat genes = self.genes skipIds = 0 geneToSym = self.geneToSym for i in range(0, len(self.genes)): geneId = genes[i] geneId, geneSym, skipIds = resolveGene(geneId, geneToSym, skipIds) logging.debug("geneId %s, geneSym %s", geneId, geneSym) if i%1000==0: logging.info("%d genes written..." % i) arr = mat.getrow(i).toarray() - #arr = arr[arr==numpy.nan]=numpy.nan if arr.ndim==2: # scipy sparse arrays have changed their entire data model and now all operations # return 2D matrices. So need to unpack it to get the array. Grrr. arr = arr[0] yield (geneId, geneSym, arr) def resolveGene(gene, geneToSym, skipIds): if "|" in gene: gene, symbol = gene.split("|") else: if geneToSym is None: symbol = gene else: if gene.startswith("EN"): gene = gene.split(".")[0] @@ -2061,50 +2060,55 @@ - len(descStr) bytes: the descriptive string descStr - array of n 4-byte floats (n = number of cells) or 4-byte unsigned ints """ geneDesc = str(geneDesc) # make sure no unicode geneIdLen = struct.pack("<H", len(geneDesc)) # on cortex-dev, numpy was around 30% faster. Not a huge difference. if numpyLoaded: if matType=="float": exprArr = exprArr.astype("float32") elif matType=="int": exprArr = exprArr.astype("uint32") else: assert(False) # internal error exprStr = exprArr.tobytes() + #minVal = np.amin(exprArr[~np.isneginf(exprArr)]) + exprArr[np.isnan(exprArr)] = FLOATNAN minVal = np.amin(exprArr) + # cortex-dev-splicing/psi has NAN values in the mtx file else: if matType=="float": arrType = "f" elif matType=="int" or matType=="forceInt": arrType = "L" else: assert(False) # internal error # if as too-old numpy version is loaded isNumpy is false, but the type may # still be a numpy array if we loaded from MTX -> force to a list if str(type(exprArr))=="<type 'numpy.ndarray'>": exprArr = exprArr.tolist()[0] + exprArr = [FLOATNAN if math.isnan(x) else x for x in exprArr] # Python 3.9 removed tostring() if sys.version_info >= (3, 2): exprStr = array.array(arrType, exprArr).tobytes() else: exprStr = array.array(arrType, exprArr).tostring() + #minVal = min([x for x in exprArr if not (math.isinf(x) and x < 0)]) # this is super slow... minVal = min(exprArr) if isPy3: geneStr = geneIdLen+bytes(geneDesc, encoding="ascii")+exprStr else: geneStr = geneIdLen+geneDesc+exprStr geneCompr = zlib.compress(geneStr) fact = float(len(geneCompr)) / len(geneStr) logging.debug("raw - compression factor of %s: %f, before %d, after %d"% (geneDesc, fact, len(geneStr), len(geneCompr))) return geneCompr, minVal def indexAtacOffsetsByChrom(exprIndex): """ given a dict with name -> (offset, len) and name being a string of chrom:start-end, @@ -2242,31 +2246,31 @@ else: if atacChromCount > 100: errAbort("There are more than 100 genes that look like a chrom_start_end range but the atacSearch cellbrowser.conf" " is not set. Please add this option or contact us if you are confused about this error.") if highCount==0: logging.warn("No single value in the matrix is > 100. It looks like this " "matrix has been log'ed. Our recommendation for visual inspection is to not transform matrices") if len(exprIndex)==0: errAbort("No genes from the expression matrix could be mapped to symbols." "Are you sure these are Ensembl IDs? Adapt geneIdType in cellbrowser.conf.") # keep a flag so the client later can figure out if the expression matrix contains any negative values # this is important for handling the 0-value - exprIndex["_range"] = (int(allMin),0) + exprIndex["_range"] = (float(allMin),0) # float() in case it is -inf as a np.float32, must be native Python float, not numpy logging.info("Global minimum in matrix is: %f" % allMin) jsonOfh = open(jsonFname, "w") json.dump(exprIndex, jsonOfh) jsonOfh.close() jsonOfh = open(discretJsonFname, "w") json.dump(discretIndex, jsonOfh) jsonOfh.close() renameFile(tmpFname, binFname) renameFile(discretTmp, discretBinFname) return matType @@ -5736,31 +5740,30 @@ collInfo["abstract"] = """This is a local Cell Browser installation, without dataset hierarchies. Please select a dataset from the list on the left.<p> <p> See the documentation on <a target=_blank href="https://cellbrowser.readthedocs.io/collections.html"> dataset hierarchies</a>. """ summInfo = summarizeDatasets(datasets) collInfo["datasets"] = summInfo outFname = join(outDir, "dataset.json") writeJson(collInfo, outFname) def checkDsCase(inConfFname, relPath, inConfig): """ relPath should not be uppercase for top-level datasets at UCSC, as we use the hostname part """ - if not inConfFname.endswith(".cellbrowser.conf") and not inConfFname.endswith(".cellbrowser"): if not "/" in relPath and getConfig("onlyLower", False) and \ "name" in inConfig and inConfig["name"].isupper(): errAbort("dataset name or directory name should not contain uppercase characters, as these do not work " "if the dataset name is specified in the URL hostname itself (e.g. cortex-dev.cells.ucsc.edu)") def build(confFnames, outDir, port=None, doDebug=False, devMode=False, redo=None): " build browser from config files confFnames into directory outDir and serve on port " outDir = resolveOutDir(outDir) if outDir=="" or outDir==None: outDir = defOutDir outDir = expanduser(outDir) if not isdir(outDir): logging.warn("The directory %s does not exist. Making a new directory now." % (outDir))