Should PAR locus on the Y chromosome be removed in RSEM pipeline?

[wangshun1121]

Hello Dr Vivian:

I developed a raw pipeline following your instruction. I checked the RSEM gene expression results on Xena:

https://xenabrowser.net/datapages/?dataset=tcga_RSEM_gene_fpkm&host=https%3A%2F%2Ftoil.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443

You have 50 more gene IDs than mine. Then I checked gene IDs between yours and mine, and found that all these 50 gene IDs are begin with ENSGR. What's more, these genes are all from PAR locus on the Y chromosome.

Your results indicate that you didn't remove these PAR locus in RSEM pipeline as you did in Kallisto. In my opinion, these genes from PAR locus should also be removed in RSEM pipelines.

[wangshun1121]

The 50 genes from PAR locus are here, from your gene table

id	gene	chrom	chromStart	chromEnd	strand
ENSGR0000228572.6	LL0YNC03-29C1.1	chrY	253743	255091	+
ENSGR0000182378.12	PLCXD1	chrY	276322	303356	+
ENSGR0000178605.12	GTPBP6	chrY	304529	318819	-
ENSGR0000226179.5	LINC00685	chrY	320990	321851	+
ENSGR0000167393.16	PPP2R3B	chrY	333963	386955	-
ENSGR0000281849.2	RP13-465B17.4	chrY	386980	405579	+
ENSGR0000275287.4	Metazoa_SRP	chrY	388100	388389	-
ENSGR0000280767.2	RP13-465B17.5	chrY	419157	421980	+
ENSGR0000234958.5	FABP5P13	chrY	523775	524102	-
ENSGR0000229232.5	KRT18P53	chrY	545236	545352	-
ENSGR0000185960.12	SHOX	chrY	624344	659411	+
ENSGR0000237531.5	RP11-309M23.1	chrY	990221	994365	+
ENSGR0000225661.6	RPL14P5	chrY	1008503	1010101	-
ENSGR0000205755.10	CRLF2	chrY	1187549	1212750	-
ENSGR0000198223.14	CSF2RA	chrY	1268800	1310381	+
ENSGR0000264510.5	BX649553.3	chrY	1291755	1291828	+
ENSGR0000264819.5	BX649553.4	chrY	1292094	1292167	+
ENSGR0000263980.5	BX649553.2	chrY	1293615	1293689	+
ENSGR0000265658.5	MIR3690	chrY	1293918	1293992	+
ENSGR0000263835.5	BX649553.1	chrY	1294132	1294206	+
ENSGR0000223274.5	RNA5SP498	chrY	1300256	1300375	-
ENSGR0000185291.10	IL3RA	chrY	1336616	1382689	+
ENSGR0000169100.12	SLC25A6	chrY	1386152	1392724	-
ENSGR0000236871.6	LINC00106	chrY	1396427	1399402	+
ENSGR0000236017.7	ASMTL-AS1	chrY	1401769	1414028	+
ENSGR0000169093.14	ASMTL	chrY	1403139	1453762	-
ENSGR0000182162.9	P2RY8	chrY	1462572	1537107	-
ENSGR0000197976.10	AKAP17A	chrY	1591593	1602514	+
ENSGR0000196433.11	ASMT	chrY	1615001	1643081	+
ENSGR0000223511.5	RP13-297E16.4	chrY	1732584	1755985	+
ENSGR0000234622.5	RP13-297E16.5	chrY	1767347	1768776	+
ENSGR0000169084.12	DHRSX	chrY	2219516	2502805	-
ENSGR0000223571.5	DHRSX-IT1	chrY	2334295	2336410	-
ENSGR0000214717.9	ZBED1	chrY	2486414	2500967	-
ENSGR0000277120.4	MIR6089	chrY	2609191	2609254	+
ENSGR0000223773.6	CD99P1	chrY	2609348	2657229	+
ENSGR0000230542.5	LINC00102	chrY	2612988	2615347	-
ENSGR0000002586.17	CD99	chrY	2691179	2741309	+
ENSGR0000168939.10	SPRY3	chrY	56954332	56968979	+
ENSGR0000237801.5	AMD1P2	chrY	57015105	57016096	-
ENSGR0000237040.5	DPH3P2	chrY	57062156	57062405	+
ENSGR0000124333.14	VAMP7	chrY	57067813	57130289	+
ENSGR0000228410.5	TCEB1P24	chrY	57165512	57165845	-
ENSGR0000223484.6	TRPC6P	chrY	57171890	57172769	-
ENSGR0000124334.16	IL9R	chrY	57184101	57197337	+
ENSGR0000270726.5	AJ271736.10	chrY	57190738	57208756	+
ENSGR0000185203.11	WASIR1	chrY	57201143	57203357	-
ENSGR0000182484.14	WASH6P	chrY	57207346	57212230	+
ENSGR0000276543.4	AJ271736.1	chrY	57209151	57209218	+
ENSGR0000227159.7	DDX11L16	chrY	57212184	57214397	-

[wangshun1121]

My suggestion: These genes should be paired with their homogenous genes on chrX. Then expression value of these genes should be added to their homogenous genes, and then remove these genes on chrY from meta table of gene expression.

[jvivian]

Dear @wangshun1121 ,

My suggestion: These genes should be paired with their homogenous genes on chrX. Then expression value of these genes should be added to their homogenous genes, and then remove these genes on chrY from meta table of gene expression.

Thank you immensely for this write-up and corresponding table. Due to demands that the workflow produce deterministic expression results that match the values on Xena, I have to freeze the default workflow inputs, but I've wanted to start hosting new sets of inputs as new Gencode annotations get released and the issues you've brought up are important points to consider.

I'll keep this issue open until I've properly addressed it. Please let me know if you have any other suggestions or improvements I can make, they're greatly appreciated.